WEBVTT

1
00:00:00.080 --> 00:00:03.240
<v Speaker 1>Welcome to the deep dive. Today, we're tackling a really

2
00:00:04.160 --> 00:00:08.000
<v Speaker 1>interesting challenge. How do we actually build the cutting edge

3
00:00:08.039 --> 00:00:13.039
<v Speaker 1>tech that's powering modern retail, you know, things like RFID tracking,

4
00:00:13.400 --> 00:00:16.600
<v Speaker 1>automated software agents, but do it in a way that's

5
00:00:16.960 --> 00:00:18.600
<v Speaker 1>well legal by design.

6
00:00:18.800 --> 00:00:21.519
<v Speaker 2>Yeah, that's the core idea our guide. Here is the

7
00:00:21.559 --> 00:00:26.359
<v Speaker 2>book Legal Programming, Designing Legally Compliant RFID and Software Agent

8
00:00:26.440 --> 00:00:29.320
<v Speaker 2>Architectures for Retail Processes and Beyond.

9
00:00:29.120 --> 00:00:31.920
<v Speaker 1>By Brian Subirana and Malcolm Bain. It's part of this

10
00:00:32.039 --> 00:00:34.960
<v Speaker 1>bigger series, the Integrated Series and Information.

11
00:00:34.640 --> 00:00:37.280
<v Speaker 2>System exactly in this book, it really sits at the

12
00:00:37.280 --> 00:00:40.759
<v Speaker 2>heart of a crucial discussion, how do we bring legal thinking,

13
00:00:41.159 --> 00:00:43.880
<v Speaker 2>you know, right into the initial design phase.

14
00:00:43.719 --> 00:00:46.119
<v Speaker 1>Instead of building something and then uh oh, trying to

15
00:00:46.119 --> 00:00:47.439
<v Speaker 1>figure out if it breaks the law.

16
00:00:47.359 --> 00:00:51.159
<v Speaker 2>Later, precisely baking in compliance from the get go, which

17
00:00:51.159 --> 00:00:53.880
<v Speaker 2>is becoming so important as we see more automation like

18
00:00:54.240 --> 00:00:57.799
<v Speaker 2>RFID and inventory maybe even autonomous shopping agents down the line. Yeah.

19
00:00:57.799 --> 00:01:00.280
<v Speaker 1>Absolutely, And the scope here is pretty wide, isn't. The

20
00:01:00.280 --> 00:01:04.400
<v Speaker 1>book covers contracts, intellectual property, consumer protection, privacy.

21
00:01:04.280 --> 00:01:07.079
<v Speaker 2>All seen through this lens of new retail tech.

22
00:01:07.040 --> 00:01:10.280
<v Speaker 1>Right, and they use this neat framing device thinking about

23
00:01:10.319 --> 00:01:13.400
<v Speaker 1>society through its retail. Like looking at Passa San Marco

24
00:01:13.439 --> 00:01:17.159
<v Speaker 1>ages ago told you about Venice. Today they say, our

25
00:01:17.280 --> 00:01:21.159
<v Speaker 1>store trolleyview gives us similar insights into our well increasingly

26
00:01:21.239 --> 00:01:22.760
<v Speaker 1>digital and automated world.

27
00:01:22.920 --> 00:01:27.040
<v Speaker 2>And that store trolleyview really highlights something fundamental. These simple

28
00:01:27.120 --> 00:01:31.680
<v Speaker 2>acts browsing, buying, they're wrapped up in this complex web

29
00:01:31.840 --> 00:01:33.439
<v Speaker 2>of legal and social.

30
00:01:33.120 --> 00:01:35.439
<v Speaker 1>Rules which technology is now kind of shaking up.

31
00:01:35.599 --> 00:01:38.680
<v Speaker 2>Exactly. It forces us to rethink how those rules apply

32
00:01:38.959 --> 00:01:40.239
<v Speaker 2>when things get more automated.

33
00:01:40.439 --> 00:01:43.680
<v Speaker 1>Okay, so our mission for this deep dive really get

34
00:01:43.760 --> 00:01:46.280
<v Speaker 1>to grips with the specific legal risks and maybe some

35
00:01:46.319 --> 00:01:50.400
<v Speaker 1>solutions when we mix online retail, RFID and especially these

36
00:01:50.400 --> 00:01:52.439
<v Speaker 1>software agents handling transactions.

37
00:01:52.480 --> 00:01:54.799
<v Speaker 2>And the book points out this core tension right away,

38
00:01:54.920 --> 00:01:57.879
<v Speaker 2>which is that a lot of online stuff happening now

39
00:01:57.959 --> 00:02:02.760
<v Speaker 2>might already be legally well questionable, and moving towards agents

40
00:02:02.760 --> 00:02:06.799
<v Speaker 2>and RFID environments that could seriously amplify the risks.

41
00:02:06.760 --> 00:02:11.439
<v Speaker 1>Because these agent systems they promise efficiency, Yeah, but they're

42
00:02:11.479 --> 00:02:16.159
<v Speaker 1>operating under laws made for people talking to people essentially, right, That's.

43
00:02:16.000 --> 00:02:20.719
<v Speaker 2>The fundamental challenge bridging the gap between this fast, autonomous

44
00:02:20.759 --> 00:02:24.840
<v Speaker 2>tech and this sort of deliberate, consent based world of law.

45
00:02:25.000 --> 00:02:27.599
<v Speaker 1>And that's where this process view comes in. The idea

46
00:02:27.639 --> 00:02:31.280
<v Speaker 1>is focusing on the processes themselves might help align tech

47
00:02:31.319 --> 00:02:31.879
<v Speaker 1>with law.

48
00:02:32.319 --> 00:02:36.159
<v Speaker 2>That's the proposal. So maybe we start with contracts. What

49
00:02:36.199 --> 00:02:38.800
<v Speaker 2>does the book say about how they're formed legally, especially

50
00:02:38.840 --> 00:02:43.919
<v Speaker 2>online where things feel so frictionless. Okay, contracts, well, traditionally

51
00:02:43.919 --> 00:02:47.879
<v Speaker 2>you need an offer, acceptance and clear consent, simple enough

52
00:02:47.919 --> 00:02:51.039
<v Speaker 2>on paper. What but how those are interpreted that can

53
00:02:51.120 --> 00:02:54.400
<v Speaker 2>vary quite a bit legally jurisdiction to jurisdiction. Even clicking

54
00:02:54.439 --> 00:02:55.919
<v Speaker 2>I agree online?

55
00:02:56.000 --> 00:02:56.960
<v Speaker 1>Yeah, what about that?

56
00:02:57.439 --> 00:03:00.560
<v Speaker 2>In one place that might seal the deal acceptance. Elsewhere

57
00:03:00.719 --> 00:03:02.360
<v Speaker 2>it might just be you making an offer that the

58
00:03:02.360 --> 00:03:05.759
<v Speaker 2>website still needs to formally accept. It's not always clear cut, and.

59
00:03:05.719 --> 00:03:07.879
<v Speaker 1>The book flags a lot of potential issues with online

60
00:03:07.879 --> 00:03:10.719
<v Speaker 1>contracts right, things like terms you can't negotiate.

61
00:03:10.560 --> 00:03:14.919
<v Speaker 2>Unilateral terms yeah, or product descriptions missing key details, hidden costs,

62
00:03:14.960 --> 00:03:16.080
<v Speaker 2>popping up late.

63
00:03:16.159 --> 00:03:19.879
<v Speaker 1>And maybe no solid record of what was actually agreed exactly.

64
00:03:20.400 --> 00:03:23.280
<v Speaker 2>The EU e Commerce Directive tries to tackle some of this,

65
00:03:23.400 --> 00:03:25.400
<v Speaker 2>but it's still a complex landscape.

66
00:03:25.439 --> 00:03:28.400
<v Speaker 1>And then we layer on more tech. You mentioned IPv six.

67
00:03:28.360 --> 00:03:31.039
<v Speaker 2>Right, the next Internet protocol. The key thing there is

68
00:03:31.080 --> 00:03:34.639
<v Speaker 2>it can assign unique digital IDs to pretty much every device,

69
00:03:34.680 --> 00:03:40.120
<v Speaker 2>which means tracking, potentially very fine grain tracking. Now there

70
00:03:40.159 --> 00:03:44.439
<v Speaker 2>are privacy extensions being discussed, but that underlying ability for

71
00:03:44.479 --> 00:03:48.719
<v Speaker 2>embedded ID, big privacy implications could change how we think

72
00:03:48.759 --> 00:03:49.919
<v Speaker 2>about online consent.

73
00:03:50.120 --> 00:03:53.599
<v Speaker 1>And then there's RFID, this Internet of things where items

74
00:03:53.639 --> 00:03:56.479
<v Speaker 1>have unique IDs can be tracked, trigger actions.

75
00:03:56.159 --> 00:03:58.439
<v Speaker 2>Super efficient for retail inventory management.

76
00:03:58.479 --> 00:04:01.280
<v Speaker 1>All that, but as the book points out, huge privacy

77
00:04:01.319 --> 00:04:04.319
<v Speaker 1>worries too. Imagine every single thing you own broadcasting its

78
00:04:04.360 --> 00:04:05.840
<v Speaker 1>location or history.

79
00:04:06.479 --> 00:04:10.080
<v Speaker 2>And all this feeds into the legal uncertainty because the Internet, well,

80
00:04:10.199 --> 00:04:11.199
<v Speaker 2>it doesn't care about.

81
00:04:10.960 --> 00:04:15.000
<v Speaker 1>Borders, different countries, different rules. Privacy is a big example.

82
00:04:15.280 --> 00:04:18.720
<v Speaker 2>Definitely, the US is moving towards more legislation, but it's

83
00:04:18.759 --> 00:04:23.480
<v Speaker 2>still different from say EU's GDPR approach, which came after

84
00:04:23.519 --> 00:04:27.920
<v Speaker 2>the book, but shows the trend. Navigating these differences is tough.

85
00:04:28.720 --> 00:04:32.120
<v Speaker 1>So the EU tried this regulation at origin thing, the.

86
00:04:32.160 --> 00:04:34.879
<v Speaker 2>E commerce directive. Yeah, basically you follow the rules where

87
00:04:34.920 --> 00:04:37.720
<v Speaker 2>you're based. Yeah, but even that gets complicated, especially with

88
00:04:37.720 --> 00:04:39.279
<v Speaker 2>things like VAT It's.

89
00:04:39.240 --> 00:04:43.120
<v Speaker 1>Wild how deep the legal rabbit holes go. Like domain names.

90
00:04:43.160 --> 00:04:46.000
<v Speaker 1>The book mentions dot com domains in Virginia.

91
00:04:46.160 --> 00:04:49.040
<v Speaker 2>Yeah, the fact that legal action against the domain itself

92
00:04:49.199 --> 00:04:51.519
<v Speaker 2>can happen in Virginia because the root server is there.

93
00:04:52.079 --> 00:04:55.279
<v Speaker 2>It means jurisdiction could potentially land there no matter where

94
00:04:55.279 --> 00:04:55.800
<v Speaker 2>the owner is.

95
00:04:55.920 --> 00:04:56.720
<v Speaker 1>Yeah, okay.

96
00:04:56.759 --> 00:04:58.879
<v Speaker 2>It really just highlights that the law for e commerce

97
00:04:58.959 --> 00:05:01.120
<v Speaker 2>is still, you know, evol It feels a bit like

98
00:05:01.120 --> 00:05:03.879
<v Speaker 2>a patchwork sometimes compared to traditional commerce.

99
00:05:03.519 --> 00:05:09.600
<v Speaker 1>Law, and that uncertainty fuels consumer worries about fraud, data breaches.

100
00:05:10.120 --> 00:05:13.519
<v Speaker 2>Leads people to maybe prefer buying within their own country

101
00:05:13.519 --> 00:05:16.319
<v Speaker 2>where they feel the legal ground is firmer. Though things

102
00:05:16.399 --> 00:05:19.360
<v Speaker 2>like BBB Online are trying to help with cross border disputes.

103
00:05:19.639 --> 00:05:22.199
<v Speaker 1>So amidst all this, the book maps out how e

104
00:05:22.240 --> 00:05:24.839
<v Speaker 1>commerce models evolved from EDI.

105
00:05:24.879 --> 00:05:29.480
<v Speaker 2>Electronic data interchange, yeah, very structured, then electronic marketplaces and

106
00:05:29.560 --> 00:05:32.920
<v Speaker 2>now these complex transaction streams with loads of intermediaries.

107
00:05:33.120 --> 00:05:36.240
<v Speaker 1>And it makes a distinction between controlled B to B

108
00:05:36.360 --> 00:05:39.439
<v Speaker 1>platforms versus open B two C website.

109
00:05:39.079 --> 00:05:42.360
<v Speaker 2>Right hierarchical versus market based. They need different trust mechanisms.

110
00:05:42.519 --> 00:05:45.040
<v Speaker 2>B to B often has pre agreed rules, B two

111
00:05:45.079 --> 00:05:47.879
<v Speaker 2>C needs more on the fly trust building.

112
00:05:47.639 --> 00:05:49.560
<v Speaker 1>And the book suggests the way these platforms are built

113
00:05:49.600 --> 00:05:54.240
<v Speaker 1>contributes to legal issues like the build first, ask later mindset.

114
00:05:54.439 --> 00:05:57.759
<v Speaker 2>That and the sheer speed of web development prioritizing features

115
00:05:57.800 --> 00:06:01.959
<v Speaker 2>over compliance, plus business models relying heavily on data collection.

116
00:06:02.680 --> 00:06:05.439
<v Speaker 2>It can all push legal compliance down the priority list.

117
00:06:05.639 --> 00:06:08.759
<v Speaker 1>And the standard non negotiable terms don't help legally either.

118
00:06:09.000 --> 00:06:11.879
<v Speaker 2>Definitely not So now let's bring in software agents. Okay.

119
00:06:11.959 --> 00:06:15.160
<v Speaker 2>The book makes a key distinction at technical agent software

120
00:06:15.360 --> 00:06:19.360
<v Speaker 2>acting independently versus legal agency, which is about authority to

121
00:06:19.399 --> 00:06:20.199
<v Speaker 2>act for someone else.

122
00:06:20.360 --> 00:06:21.519
<v Speaker 1>Different things but related.

123
00:06:21.680 --> 00:06:24.839
<v Speaker 2>The parallels are useful. Yeah, and it mentions formal models

124
00:06:24.879 --> 00:06:29.639
<v Speaker 2>like BDI, beliefs, desires, intentions, trying to map out agent thinking,

125
00:06:30.120 --> 00:06:32.560
<v Speaker 2>which is important when you ask who's responsible later?

126
00:06:32.759 --> 00:06:36.000
<v Speaker 1>Right. Understanding the logic helps assign accountability, and the book

127
00:06:36.040 --> 00:06:40.360
<v Speaker 1>shows agents could do a lot in e commerce, filtering data, personal.

128
00:06:39.920 --> 00:06:44.680
<v Speaker 2>Assistance, negotiating deals, managing risk. It's a long list, and.

129
00:06:44.560 --> 00:06:47.800
<v Speaker 1>It connects closed agent platforms to those hierarchical models and

130
00:06:47.879 --> 00:06:50.959
<v Speaker 1>open platforms to the market based ones. Makes sense.

131
00:06:51.240 --> 00:06:55.079
<v Speaker 2>Currently agents are mostly helpers right search engines, shopping assistants

132
00:06:55.360 --> 00:06:56.920
<v Speaker 2>still under user control.

133
00:06:56.680 --> 00:06:59.839
<v Speaker 1>But the book sees more intelligent agents coming ones that

134
00:06:59.839 --> 00:07:02.480
<v Speaker 1>can initiate deals on their own learn.

135
00:07:02.519 --> 00:07:05.920
<v Speaker 2>Yeah, think automated warehouse restocking or an agent doing your

136
00:07:05.920 --> 00:07:07.600
<v Speaker 2>weekly grocery shop autonomously.

137
00:07:07.759 --> 00:07:12.279
<v Speaker 1>And RFID amplifies this because physical items feed data directly

138
00:07:12.360 --> 00:07:13.839
<v Speaker 1>to these agents exactly.

139
00:07:14.079 --> 00:07:17.920
<v Speaker 2>It creates a much more dynamic, potentially autonomous system. The

140
00:07:17.920 --> 00:07:21.360
<v Speaker 2>book paints this picture with agents for contracts, data management,

141
00:07:21.639 --> 00:07:24.040
<v Speaker 2>user profiling, monitoring, even consumer protection.

142
00:07:24.720 --> 00:07:28.600
<v Speaker 1>Very automated future sounds convenient, but the book flags the downsides.

143
00:07:28.639 --> 00:07:31.160
<v Speaker 1>Agents might help with compliance sometimes, but.

144
00:07:31.120 --> 00:07:34.759
<v Speaker 2>They also create new legal headaches, big ones. Can an

145
00:07:34.800 --> 00:07:38.000
<v Speaker 2>agent have legal personality? How do we get valid consent

146
00:07:38.319 --> 00:07:41.879
<v Speaker 2>for contracts or data use via an agent? What about

147
00:07:41.959 --> 00:07:44.560
<v Speaker 2>IP rights when agents use online content?

148
00:07:44.920 --> 00:07:51.040
<v Speaker 1>So many questions, contract formation, liability, privacy, IPR identification, who's

149
00:07:51.079 --> 00:07:55.560
<v Speaker 1>responsible for what the agent says? Monitoring mistakes? Jurisdiction too.

150
00:07:55.720 --> 00:07:59.000
<v Speaker 2>It's a long list, and legal systems are starting to notice.

151
00:07:59.199 --> 00:08:02.439
<v Speaker 2>The book mentions definitions of electronic agent appearing in laws

152
00:08:02.480 --> 00:08:05.120
<v Speaker 2>like UADA in the US and unsentral.

153
00:08:04.560 --> 00:08:08.079
<v Speaker 1>Models focusing on that automated action without human review.

154
00:08:08.160 --> 00:08:10.199
<v Speaker 2>Bit yeah, the law trying to catch up. To make

155
00:08:10.199 --> 00:08:12.800
<v Speaker 2>this concrete, The book uses a research scenario.

156
00:08:12.560 --> 00:08:14.920
<v Speaker 1>Right, the augmented shopping experience.

157
00:08:14.480 --> 00:08:17.879
<v Speaker 2>Using phones, smart cards, in store screens for personalized stuff,

158
00:08:18.160 --> 00:08:23.639
<v Speaker 2>parking directions, smart shopping lists, dietary reminders, product locations, special offers, really.

159
00:08:23.399 --> 00:08:27.240
<v Speaker 1>Bringing mobile commerce commerce into the physical store. It mentions

160
00:08:27.279 --> 00:08:32.240
<v Speaker 1>the tech making this feasible gprs, Bluetooth, Wi Fi, powerful phones, and.

161
00:08:32.159 --> 00:08:35.639
<v Speaker 2>It connects to ubiquitous computing tech fading into the background,

162
00:08:36.039 --> 00:08:40.639
<v Speaker 2>embedded everywhere, like the EUS disappearing computer initiative envisioned.

163
00:08:40.320 --> 00:08:42.919
<v Speaker 1>And they give practical examples of things going wrong, like

164
00:08:43.039 --> 00:08:45.960
<v Speaker 1>your shopping bot buying way too much stuff.

165
00:08:45.639 --> 00:08:48.399
<v Speaker 2>Or getting a location based offer you didn't really ask for. Right, then,

166
00:08:48.480 --> 00:08:53.200
<v Speaker 2>who's responsible? Did you really? Consent raises immedia legal questions.

167
00:08:52.799 --> 00:08:54.919
<v Speaker 1>And it's not just the shopper and store involved. There

168
00:08:54.960 --> 00:09:01.600
<v Speaker 1>are advisors, service providers, data repositories, logistics advertisers, the regulator,

169
00:09:02.159 --> 00:09:03.559
<v Speaker 1>lots of actors, each with.

170
00:09:03.480 --> 00:09:07.480
<v Speaker 2>Different relationships and responsibilities. It's a complex ecosystem.

171
00:09:07.039 --> 00:09:10.159
<v Speaker 1>So to analyze it, the book introduces four agent types

172
00:09:10.679 --> 00:09:15.320
<v Speaker 1>A store advertising, B, store selling, C customer buying agent,

173
00:09:15.399 --> 00:09:17.360
<v Speaker 1>and D shopping assistant.

174
00:09:16.919 --> 00:09:19.919
<v Speaker 2>Right, each with different functions, different legal angles.

175
00:09:19.559 --> 00:09:22.720
<v Speaker 1>And the core legal areas to examine our contracts again,

176
00:09:23.320 --> 00:09:27.240
<v Speaker 1>automated formation, intent, consent signatures.

177
00:09:26.759 --> 00:09:32.759
<v Speaker 2>Intellectual property access, copyright, trademarks, databases, consumer protection ads, offers,

178
00:09:32.799 --> 00:09:36.840
<v Speaker 2>delivery info, and privacy collecting, storing processing data.

179
00:09:36.879 --> 00:09:39.799
<v Speaker 1>These are the big pillars impacted by adding agents and

180
00:09:40.000 --> 00:09:43.279
<v Speaker 1>RFID building on existing Internet law, and the.

181
00:09:43.240 --> 00:09:47.120
<v Speaker 2>Focus is really on the additional issues these new technologies raise.

182
00:09:47.480 --> 00:09:51.559
<v Speaker 2>It distinguishes between observer agents just gathering.

183
00:09:51.080 --> 00:09:54.480
<v Speaker 1>Info and actor agents that actually do things acquire rights.

184
00:09:54.600 --> 00:09:55.879
<v Speaker 1>They cause more legal friction.

185
00:09:56.240 --> 00:09:59.159
<v Speaker 2>Right, The book's approach is to apply existing laws to

186
00:09:59.200 --> 00:10:02.960
<v Speaker 2>these agent process ssss, then try to generalize find higher

187
00:10:03.000 --> 00:10:06.679
<v Speaker 2>level rules or meta models, mostly using the EU framework

188
00:10:06.799 --> 00:10:10.879
<v Speaker 2>with some US comparisons, but acknowledging national laws needs specifics.

189
00:10:11.320 --> 00:10:16.360
<v Speaker 1>Okay, back to contract basics consent, offer acceptance, meeting of minds,

190
00:10:16.440 --> 00:10:18.879
<v Speaker 1>free will, intention to be bound.

191
00:10:18.759 --> 00:10:21.759
<v Speaker 2>And the booknotes potential issues like mistake or bad faith,

192
00:10:22.159 --> 00:10:25.360
<v Speaker 2>plus slight differences between common law and civil law views.

193
00:10:25.080 --> 00:10:27.480
<v Speaker 1>On intent and the need for written evidence varies, but

194
00:10:27.519 --> 00:10:28.720
<v Speaker 1>it's usually best for proof.

195
00:10:28.840 --> 00:10:32.480
<v Speaker 2>Then you have key EU directives e signatures on legal standing,

196
00:10:32.679 --> 00:10:35.720
<v Speaker 2>distance selling for consumer protection in remote deals, and the

197
00:10:35.759 --> 00:10:39.519
<v Speaker 2>e Commerce Directive being absorbed into national laws. Plus specific

198
00:10:39.559 --> 00:10:42.519
<v Speaker 2>e commerce laws elsewhere like u ada ZIN in the.

199
00:10:42.559 --> 00:10:45.480
<v Speaker 1>US, all setting the stage for how agents fit in.

200
00:10:45.840 --> 00:10:49.960
<v Speaker 1>What are the practical legal issues for agent contracts procedures?

201
00:10:50.240 --> 00:10:52.840
<v Speaker 2>How does an agent know it's a real offer? Evidence,

202
00:10:52.879 --> 00:10:57.039
<v Speaker 2>meeting and writing requirements? Digitally terms? How are they incorporated

203
00:10:57.320 --> 00:11:00.799
<v Speaker 2>understood by the user via the agent? How does an

204
00:11:00.799 --> 00:11:04.320
<v Speaker 2>agent provide a binding digital one and consumer rights info?

205
00:11:04.639 --> 00:11:07.399
<v Speaker 2>Transparency consent through an autumn time?

206
00:11:08.000 --> 00:11:12.360
<v Speaker 1>And can the agent itself have legal personality own things

207
00:11:12.799 --> 00:11:13.519
<v Speaker 1>be liable?

208
00:11:14.039 --> 00:11:17.360
<v Speaker 2>The book leans towards no, not under current law. So

209
00:11:17.440 --> 00:11:21.399
<v Speaker 2>the challenges attribution linking the agent's actions back to a person.

210
00:11:21.159 --> 00:11:24.720
<v Speaker 1>Or entity, which brings in legal agency concepts, express authority,

211
00:11:24.759 --> 00:11:26.399
<v Speaker 1>a parent authority ratification.

212
00:11:26.799 --> 00:11:31.159
<v Speaker 2>Exactly could these frameworks apply? The book suggests programming an

213
00:11:31.159 --> 00:11:34.120
<v Speaker 2>agent's identity might even help establish a parent authority.

214
00:11:34.159 --> 00:11:36.279
<v Speaker 1>What about risk if the agent messes up due to

215
00:11:36.279 --> 00:11:38.240
<v Speaker 1>bad programming or unexpected learning.

216
00:11:38.480 --> 00:11:41.480
<v Speaker 2>Keeping records of initial settings and user intent seems crucial

217
00:11:41.559 --> 00:11:45.600
<v Speaker 2>for figuring out liability and consumer protection. Laws allowing cancelation

218
00:11:45.679 --> 00:11:49.000
<v Speaker 2>of distance contracts might give consumer agent users.

219
00:11:48.679 --> 00:11:52.600
<v Speaker 1>An edge digital signatures via agents. It comes back to attribution,

220
00:11:53.000 --> 00:11:55.120
<v Speaker 1>linking it to the private keyhole.

221
00:11:54.919 --> 00:11:59.039
<v Speaker 2>Yes, and Article six of the EUE Signatures Directive stresses

222
00:11:59.080 --> 00:12:00.480
<v Speaker 2>the signatory needing control.

223
00:12:00.720 --> 00:12:03.919
<v Speaker 1>Can an agent really be under the user's soul control,

224
00:12:04.480 --> 00:12:06.960
<v Speaker 1>especially if it stores passwords or it's a modal agent

225
00:12:07.039 --> 00:12:08.080
<v Speaker 1>replicating itself?

226
00:12:08.440 --> 00:12:11.360
<v Speaker 2>Big questions? Maybe the agent has to check back with

227
00:12:11.399 --> 00:12:14.360
<v Speaker 2>the user for sensitive data. The book also notes the

228
00:12:14.360 --> 00:12:17.200
<v Speaker 2>tech security requirements for secure signature.

229
00:12:16.759 --> 00:12:20.399
<v Speaker 1>Devices and do certification authorities usually require users to keep

230
00:12:20.480 --> 00:12:22.639
<v Speaker 1>keys secret, not record them.

231
00:12:22.799 --> 00:12:26.759
<v Speaker 2>Typically yes in their policies. The book also cites Unciti

232
00:12:26.840 --> 00:12:30.639
<v Speaker 2>Troll's model law on e commerce, defining the originator and

233
00:12:30.679 --> 00:12:33.919
<v Speaker 2>when a message is deemed theirs focusing on authority.

234
00:12:33.559 --> 00:12:37.279
<v Speaker 1>And the unctral draft model Law on e contracts directly

235
00:12:37.279 --> 00:12:38.799
<v Speaker 1>tackles automated systems.

236
00:12:38.879 --> 00:12:41.679
<v Speaker 2>It does as does UETA in the US, which attributes

237
00:12:41.720 --> 00:12:44.639
<v Speaker 2>agent acts to the initiator even without knowledge, but it

238
00:12:44.679 --> 00:12:48.120
<v Speaker 2>handles human error differently from machine error. U SATA also

239
00:12:48.240 --> 00:12:52.039
<v Speaker 2>recognizes agent contracts with a reasonableness test for consent, but

240
00:12:52.120 --> 00:12:53.360
<v Speaker 2>it wasn't widely adopted.

241
00:12:53.679 --> 00:12:57.600
<v Speaker 1>Any technical fixes mentioned for validating automated contract.

242
00:12:57.200 --> 00:13:01.320
<v Speaker 2>Briefly touches on policy expressions for security, especially for intelligent

243
00:13:01.399 --> 00:13:02.320
<v Speaker 2>and mobile agents.

244
00:13:02.399 --> 00:13:06.879
<v Speaker 1>Okay Let's switch gears to intellectual property rights IPR.

245
00:13:06.840 --> 00:13:11.399
<v Speaker 2>Right, copyright, trademarks, database rights. How do they apply to agents?

246
00:13:11.440 --> 00:13:15.639
<v Speaker 1>So copyright basics protects original works fixed in a tangible

247
00:13:15.679 --> 00:13:18.360
<v Speaker 1>medium literary, artistic, etc.

248
00:13:18.919 --> 00:13:22.679
<v Speaker 2>Digital content counts, even RAM storage can be fixed enough.

249
00:13:22.879 --> 00:13:26.639
<v Speaker 2>Protection is territorial, but tentees help extend it. Computer programs

250
00:13:26.679 --> 00:13:29.759
<v Speaker 2>are usually literary works, but it's the expression, not the

251
00:13:29.840 --> 00:13:34.240
<v Speaker 2>underlying idea or algorithm that's protected. Though look and feel

252
00:13:34.600 --> 00:13:36.679
<v Speaker 2>sometimes gets protection, especially.

253
00:13:36.279 --> 00:13:39.279
<v Speaker 1>In the US, and database rights in the EU. That's different.

254
00:13:39.519 --> 00:13:43.159
<v Speaker 2>Yeah, two tiers copyright if the structure is an original creation,

255
00:13:43.399 --> 00:13:46.720
<v Speaker 2>and this separate suey generous right if there's substantial investment.

256
00:13:47.039 --> 00:13:51.360
<v Speaker 2>Protecting against extracting big chunks applies even to web pages

257
00:13:51.399 --> 00:13:52.679
<v Speaker 2>computer generated lists.

258
00:13:52.960 --> 00:13:55.679
<v Speaker 1>The book has a table Table three two showing IPR

259
00:13:55.720 --> 00:13:59.879
<v Speaker 1>issues for agency's processes searching, accessing databases, storing presenting info.

260
00:14:00.320 --> 00:14:02.200
<v Speaker 1>Accessing online databases seems key.

261
00:14:02.360 --> 00:14:06.639
<v Speaker 2>That's often the frux. This section details those compliance problems,

262
00:14:07.039 --> 00:14:10.879
<v Speaker 2>looks at exemptions and touches on protecting agent created results.

263
00:14:11.679 --> 00:14:15.759
<v Speaker 2>A big concern is agents infringing by displaying copyrighted text

264
00:14:15.840 --> 00:14:17.000
<v Speaker 2>or images from websites.

265
00:14:17.080 --> 00:14:18.879
<v Speaker 1>What about just linking to content.

266
00:14:19.200 --> 00:14:22.919
<v Speaker 2>Even temporary links made by agents could be infringements, and

267
00:14:23.000 --> 00:14:26.639
<v Speaker 2>things like framing or inlining content from other sites are

268
00:14:26.720 --> 00:14:33.240
<v Speaker 2>particularly problematic, potentially reproduction, derivative works, unfair competition. There's case

269
00:14:33.279 --> 00:14:34.320
<v Speaker 2>law on this, and.

270
00:14:34.240 --> 00:14:38.519
<v Speaker 1>Could agents create derivative works by combining info from multiple

271
00:14:38.519 --> 00:14:41.600
<v Speaker 1>sites like product comparisons with images?

272
00:14:41.639 --> 00:14:45.360
<v Speaker 2>Potentially? Yes, Then the question is liability who's responsible. In

273
00:14:45.399 --> 00:14:49.559
<v Speaker 2>the US, direct infringement usually needs knowledge intent. The DMCA

274
00:14:49.679 --> 00:14:53.440
<v Speaker 2>offers safe harbors for service providers under certain conditions.

275
00:14:52.919 --> 00:14:55.799
<v Speaker 1>And the EU e Commerce Directive has similar exemptions for

276
00:14:56.000 --> 00:14:57.639
<v Speaker 1>conduits caching, hosting.

277
00:14:57.879 --> 00:15:01.000
<v Speaker 2>Yes, if they lack knowledge act fast on takedown notices,

278
00:15:01.080 --> 00:15:02.799
<v Speaker 2>but it's maybe harder to claim if the agent is

279
00:15:02.840 --> 00:15:06.039
<v Speaker 2>actively modifying or extracting data, not just passing it through.

280
00:15:06.159 --> 00:15:08.840
<v Speaker 1>Are there any copyright exemptions that might help agents?

281
00:15:09.120 --> 00:15:13.399
<v Speaker 2>The EU Copyright Directive exempts temporary incidental reproductions if they're

282
00:15:13.440 --> 00:15:16.600
<v Speaker 2>essential to a tech process and have no economic value themselves.

283
00:15:17.039 --> 00:15:18.200
<v Speaker 2>Article five point one.

284
00:15:18.360 --> 00:15:20.200
<v Speaker 1>Could that cover agent caching of.

285
00:15:20.159 --> 00:15:23.240
<v Speaker 2>Ads, for example, Possibly, but it depends if the caching

286
00:15:23.320 --> 00:15:27.240
<v Speaker 2>is truly transient or more permanent. Other exemptions like private

287
00:15:27.320 --> 00:15:29.759
<v Speaker 2>use education probably don't apply much here.

288
00:15:29.919 --> 00:15:33.080
<v Speaker 1>What about the e Commerce Directive's rules on no general

289
00:15:33.120 --> 00:15:37.759
<v Speaker 1>monitoring obligation and the notice and takedown for hosts could.

290
00:15:37.600 --> 00:15:40.360
<v Speaker 2>Be relevant if an agent service provider stores data collected

291
00:15:40.360 --> 00:15:44.080
<v Speaker 2>by agents like agent D relying on implied consent from

292
00:15:44.080 --> 00:15:46.720
<v Speaker 2>websites is tough, especially for commercial.

293
00:15:46.399 --> 00:15:49.159
<v Speaker 1>Bots and protecting the agent's own output less.

294
00:15:48.960 --> 00:15:52.840
<v Speaker 2>Focus here, but important Overall. Digital content like web pages

295
00:15:52.879 --> 00:15:56.000
<v Speaker 2>can get IPR in various ways artistic work, database rights, etc.

296
00:15:56.639 --> 00:15:59.679
<v Speaker 1>So moving to solutions for IPR compliance, it's a big

297
00:15:59.679 --> 00:16:03.759
<v Speaker 1>debate generally right copyright term open source drms.

298
00:16:03.399 --> 00:16:06.720
<v Speaker 2>Digital rights management systems. Yeah, the book mentions first generation

299
00:16:06.840 --> 00:16:11.360
<v Speaker 2>tech protections like robots, dot txt files, website owners signaling

300
00:16:11.399 --> 00:16:15.279
<v Speaker 2>preferences to bots, but they're just netiquete, not legally binding.

301
00:16:15.480 --> 00:16:21.279
<v Speaker 1>Could we see standardized dialogues agents negotiating access with website drms?

302
00:16:21.279 --> 00:16:24.799
<v Speaker 2>Potentially? Maybe not common now for general web data, but

303
00:16:25.120 --> 00:16:28.519
<v Speaker 2>could be. But there's a worry drms might be too blunt,

304
00:16:29.000 --> 00:16:33.039
<v Speaker 2>ignore fair use, create issues for users whose agents infringe

305
00:16:33.080 --> 00:16:36.960
<v Speaker 2>without them knowing. Current drms often lack flexibility.

306
00:16:36.399 --> 00:16:39.759
<v Speaker 1>So a key problem is agent autonomy versus license restrictions.

307
00:16:40.080 --> 00:16:43.320
<v Speaker 1>Agents might not understand copyright notices like humans do.

308
00:16:43.559 --> 00:16:47.080
<v Speaker 2>Exactly. And again, liability attribution is tricky. Who's the user

309
00:16:47.159 --> 00:16:51.120
<v Speaker 2>or custodian store shopper who has control could be joint

310
00:16:51.120 --> 00:16:54.200
<v Speaker 2>liability and exemptions might have played differently depending on who

311
00:16:54.279 --> 00:16:54.600
<v Speaker 2>runs the.

312
00:16:54.600 --> 00:16:56.840
<v Speaker 1>Agent, and identifying the agent is hard to makes it

313
00:16:56.879 --> 00:16:58.960
<v Speaker 1>tough for rights holders breeds, mistrust.

314
00:16:58.679 --> 00:17:02.919
<v Speaker 2>Right dynamic ips asian platforms. Identification isn't easy unless it's

315
00:17:02.919 --> 00:17:06.079
<v Speaker 2>built in which raises privacy flags. Content owners might block

316
00:17:06.160 --> 00:17:09.920
<v Speaker 2>unknown agents or demand trusted ID systems, eas trust frameworks,

317
00:17:10.079 --> 00:17:11.279
<v Speaker 2>maybe agent certification.

318
00:17:11.640 --> 00:17:15.160
<v Speaker 1>Content owners struggle to know when infringement happened, who did it.

319
00:17:15.599 --> 00:17:19.759
<v Speaker 1>Current weblogs and IPv four aren't always enough. Legal certainty

320
00:17:19.839 --> 00:17:24.279
<v Speaker 1>might have to wait for case. Law needs interoperability, release control,

321
00:17:24.440 --> 00:17:26.480
<v Speaker 1>persistence for these agents.

322
00:17:26.319 --> 00:17:30.079
<v Speaker 2>Which brings us back to designing systems differently. Higher level

323
00:17:30.119 --> 00:17:34.039
<v Speaker 2>models integrating legal issues into architecture and process models right

324
00:17:34.079 --> 00:17:36.000
<v Speaker 2>from the start embed compliance.

325
00:17:36.240 --> 00:17:39.079
<v Speaker 1>If legal rules can be modeled with standard languages, you

326
00:17:39.160 --> 00:17:41.720
<v Speaker 1>apply that model to the business process model to get

327
00:17:41.720 --> 00:17:43.119
<v Speaker 1>a compliant architecture.

328
00:17:43.240 --> 00:17:47.599
<v Speaker 2>That's the idea. Where law allows negotiation like access consent,

329
00:17:47.880 --> 00:17:52.559
<v Speaker 2>maybe standardized protocols, where it mandates procedures like specific consent forms,

330
00:17:52.799 --> 00:17:54.400
<v Speaker 2>build those in as constraints.

331
00:17:54.480 --> 00:17:56.880
<v Speaker 1>Okay, let's tackle consumer protection now, all right.

332
00:17:57.039 --> 00:18:00.440
<v Speaker 2>The book outlines the EU framework, lots of harmonization, but

333
00:18:00.559 --> 00:18:04.599
<v Speaker 2>still national laws implementing directives. The analysis looks at risks

334
00:18:04.599 --> 00:18:07.519
<v Speaker 2>for retailers and consumers from agents in the research scenario.

335
00:18:07.680 --> 00:18:10.400
<v Speaker 1>It starts with some history EC Treaty Art. One hundred

336
00:18:10.400 --> 00:18:14.359
<v Speaker 1>and fifty three initiatives leading to the Unfair Commercial Practices Directive,

337
00:18:14.720 --> 00:18:19.680
<v Speaker 1>then key directives distant selling e commerce. Unfair Commercial Practices and.

338
00:18:19.799 --> 00:18:23.519
<v Speaker 2>Specific articles in the e Commerce Directive are crucial. General

339
00:18:23.559 --> 00:18:29.240
<v Speaker 2>info requirements Art five Rules on Commercial Communications ADS basically

340
00:18:29.640 --> 00:18:33.000
<v Speaker 2>contracting info and those intermediary liability exceptions No.

341
00:18:33.240 --> 00:18:36.559
<v Speaker 1>Art Five means all service providers, including retailers using agents

342
00:18:36.559 --> 00:18:41.799
<v Speaker 1>need to provide clear details name, address, contact, VAT, pricing.

343
00:18:41.559 --> 00:18:45.440
<v Speaker 2>Yes, easily and permanently accessible. The book also mentions self

344
00:18:45.480 --> 00:18:49.079
<v Speaker 2>regulatory codes in some places and other laws like tort

345
00:18:49.200 --> 00:18:52.119
<v Speaker 2>law or trade description laws. Though the focus is consumer

346
00:18:52.160 --> 00:18:53.319
<v Speaker 2>protection directly hit.

347
00:18:53.200 --> 00:18:56.319
<v Speaker 1>By agents, and it acknowledges that consumer reluctance about e

348
00:18:56.359 --> 00:18:58.680
<v Speaker 1>commerce is still a thing. Lack of confidence.

349
00:18:58.759 --> 00:19:01.599
<v Speaker 2>Yeah, studies show it. A legal framework helps, but agents

350
00:19:01.640 --> 00:19:05.440
<v Speaker 2>add specific worries sum are technical fixes. Data protection clear info.

351
00:19:05.640 --> 00:19:09.160
<v Speaker 2>Others are gray areas. The scope here is agents interacting

352
00:19:09.200 --> 00:19:12.599
<v Speaker 2>between consumers and merchion stores providers, not the supply chain

353
00:19:12.680 --> 00:19:13.559
<v Speaker 2>or the goods themselves.

354
00:19:13.799 --> 00:19:16.559
<v Speaker 1>Table four to two maps issues in the pre contractual

355
00:19:16.559 --> 00:19:21.880
<v Speaker 1>info ads and contractual feeses. So pre contractual if stores

356
00:19:21.960 --> 00:19:25.559
<v Speaker 1>contact consumers via agents, what are the info obligations? Are

357
00:19:25.599 --> 00:19:29.839
<v Speaker 1>these information society services? Do distant selling rules apply? What

358
00:19:29.920 --> 00:19:31.839
<v Speaker 1>about consent for cold calls via agent?

359
00:19:32.079 --> 00:19:35.519
<v Speaker 2>Good questions? Does Article ten of distance selling apply to

360
00:19:35.680 --> 00:19:40.480
<v Speaker 2>SMS or agent messages? Are they emails or equivalent? Individual communications?

361
00:19:40.559 --> 00:19:44.279
<v Speaker 1>And the big one directives say provide info to the consumer?

362
00:19:44.519 --> 00:19:45.920
<v Speaker 1>Is the agent the consumer?

363
00:19:46.079 --> 00:19:49.960
<v Speaker 2>That's a major issue. Transparency is key risk the consumer

364
00:19:50.000 --> 00:19:52.720
<v Speaker 2>doesn't get the info if the agent doesn't relay it properly.

365
00:19:52.960 --> 00:19:54.880
<v Speaker 2>Courts might need to interpret this for agents.

366
00:19:55.279 --> 00:19:58.880
<v Speaker 1>Pre contractual phase has two sides, reliable product service info

367
00:19:59.000 --> 00:20:02.440
<v Speaker 1>and rules on advertise commercial communication like agent A sending

368
00:20:02.480 --> 00:20:03.920
<v Speaker 1>offers when you approach a store.

369
00:20:03.680 --> 00:20:07.680
<v Speaker 2>Section right, and commercial communications cover ads other promotions. Agents

370
00:20:07.680 --> 00:20:10.599
<v Speaker 2>A and B might send unsolicited messages triggered by RFID

371
00:20:10.880 --> 00:20:14.359
<v Speaker 2>regulated by e commerce directive. Maybe distance contracts directive too.

372
00:20:14.400 --> 00:20:17.279
<v Speaker 2>If it's like automated calling needs consent.

373
00:20:17.240 --> 00:20:21.160
<v Speaker 1>Contractual phase. Agents B and C offering goods via mobile

374
00:20:21.160 --> 00:20:24.960
<v Speaker 1>devices need the general info from e commerce directive plus

375
00:20:25.079 --> 00:20:28.839
<v Speaker 1>extra details before contract conclusion from distance selling Art.

376
00:20:29.319 --> 00:20:32.400
<v Speaker 2>Ten. And if the sale isn't an individual communication, you

377
00:20:32.440 --> 00:20:35.480
<v Speaker 2>still need to inform about contract steps, codes of conduct,

378
00:20:35.880 --> 00:20:39.759
<v Speaker 2>error correction, et cetera. Maybe argue personalized agent messages are

379
00:20:39.799 --> 00:20:45.160
<v Speaker 2>equivalent to emails. Crucially, contract terms must be accessible, storable,

380
00:20:45.440 --> 00:20:49.799
<v Speaker 2>reproducible by the consumer. RT ten point three often unmet.

381
00:20:49.880 --> 00:20:52.200
<v Speaker 2>Now might need extra agent layers.

382
00:20:52.119 --> 00:20:57.640
<v Speaker 1>Post contractual obligations, liability for performance, transaction recording, merchant disclaimers,

383
00:20:57.720 --> 00:21:02.039
<v Speaker 1>applicable law, jurisdiction. Merchants are bound by valid agent contracts

384
00:21:02.079 --> 00:21:05.160
<v Speaker 1>like current online sales software. Performance doesn't change that, but

385
00:21:05.160 --> 00:21:06.400
<v Speaker 1>should be clear to the consumer.

386
00:21:06.759 --> 00:21:09.680
<v Speaker 2>The book also looks beyond EU directors to national laws,

387
00:21:09.839 --> 00:21:13.799
<v Speaker 2>often based on fair trading principles, mentions, varying protection levels,

388
00:21:13.839 --> 00:21:16.759
<v Speaker 2>and the EC's Green Paper aiming to boost confidence. But

389
00:21:16.839 --> 00:21:19.319
<v Speaker 2>self regulatory codes have limits due to legal diversity and

390
00:21:19.400 --> 00:21:20.440
<v Speaker 2>enforceability questions.

391
00:21:20.480 --> 00:21:23.599
<v Speaker 1>What about commercial and technical solutions, Trust seals.

392
00:21:23.599 --> 00:21:29.119
<v Speaker 2>Like BBB web Trader, the audit companies against criteria website info, ordering, privacy,

393
00:21:29.160 --> 00:21:33.640
<v Speaker 2>payments disputes important but maybe not infallible, especially on privacy

394
00:21:33.640 --> 00:21:36.000
<v Speaker 2>in the US. Insurance is another option.

395
00:21:36.279 --> 00:21:42.000
<v Speaker 1>Technical proposals, cryptography labels, smart agents, tokens, biometrics, water marks, yeah.

396
00:21:41.799 --> 00:21:44.759
<v Speaker 2>Things like SSL set tee for payments, visa a smart

397
00:21:44.759 --> 00:21:46.559
<v Speaker 2>card reader. Those set adoption was.

398
00:21:46.559 --> 00:21:50.480
<v Speaker 1>Limited regulatory developments. The EC Green Paper from two thousand

399
00:21:50.519 --> 00:21:54.640
<v Speaker 1>and one aimed for more harmonization cooperation, suggested self regulation

400
00:21:54.720 --> 00:21:58.519
<v Speaker 1>within legislation, proposed a framework directive on fear trading right.

401
00:21:58.599 --> 00:22:02.119
<v Speaker 2>Those specific laws might still be needed. Debate was harmonization

402
00:22:02.240 --> 00:22:06.279
<v Speaker 2>versus core principles mentioned to sales Promotion regulation proposal. Then

403
00:22:06.319 --> 00:22:09.119
<v Speaker 2>the two thousand and three Unfair Business Practices Directive proposal

404
00:22:09.160 --> 00:22:12.759
<v Speaker 2>banning pyramid selling undisclosed paid media had a general test

405
00:22:12.759 --> 00:22:16.640
<v Speaker 2>for unfairness, define misleading and aggressive practices. Table four to

406
00:22:16.680 --> 00:22:18.440
<v Speaker 2>three summarizes issues for agent.

407
00:22:18.240 --> 00:22:21.759
<v Speaker 1>B okay onto the Big One, privacy, fundamental right, increasing

408
00:22:21.759 --> 00:22:25.920
<v Speaker 1>concerns with tech making data collection easier, especially in e commerce, and.

409
00:22:26.000 --> 00:22:29.079
<v Speaker 2>Very relevant to the research scenario with RFID and agents

410
00:22:29.119 --> 00:22:33.720
<v Speaker 2>handling personal data. Book gives historical context OECD Guidelines nineteen

411
00:22:33.759 --> 00:22:36.200
<v Speaker 2>eighty Council of Europe Treaty one to eight nineteen eighty

412
00:22:36.200 --> 00:22:37.640
<v Speaker 2>one influencing EU law.

413
00:22:37.759 --> 00:22:40.079
<v Speaker 1>Then the key principles of the nineteen ninety five EU

414
00:22:40.200 --> 00:22:46.400
<v Speaker 1>Data Directive purpose limitation, data quality, proportionality, lawful basis, consent, contract,

415
00:22:46.559 --> 00:22:51.079
<v Speaker 1>legal need, legitimate interest, transparency, no sense of data processing

416
00:22:51.119 --> 00:22:53.079
<v Speaker 1>generally security plus.

417
00:22:52.920 --> 00:22:58.839
<v Speaker 2>Data subject rights, access, rectification, objection amisrator direct marketing, no

418
00:22:58.960 --> 00:23:03.039
<v Speaker 2>automated decisions with significant effect and some exemptions. Also the

419
00:23:03.119 --> 00:23:06.960
<v Speaker 2>international angle need for adequate protection for data transfers outside

420
00:23:06.960 --> 00:23:10.920
<v Speaker 2>EU mentioned safe harbor specific approved countries provides a summary

421
00:23:11.039 --> 00:23:12.240
<v Speaker 2>rules for the research scenario.

422
00:23:12.279 --> 00:23:15.079
<v Speaker 1>And privacy is fundamental to advanced e commerce. Right so

423
00:23:15.160 --> 00:23:18.039
<v Speaker 1>much relies on personal data, the focus shifts to specific

424
00:23:18.119 --> 00:23:22.519
<v Speaker 1>risks from agent processing Beyond general eBusiness stuff Ubiquitous computing

425
00:23:22.559 --> 00:23:25.240
<v Speaker 1>means wider, more intense data collection is possible.

426
00:23:25.680 --> 00:23:30.960
<v Speaker 2>General risks get amplified, unauthorized collection, processing, transfer profiling, lack

427
00:23:31.000 --> 00:23:34.599
<v Speaker 2>of transparency, bad security. Not the tech itself, but the

428
00:23:34.640 --> 00:23:39.200
<v Speaker 2>applications and delegation is key. You trust the agent. As

429
00:23:39.279 --> 00:23:41.680
<v Speaker 2>agents get more data, they hold more sensitive info that

430
00:23:41.720 --> 00:23:44.480
<v Speaker 2>can be revealed accidentally due to their autonomy.

431
00:23:44.720 --> 00:23:49.440
<v Speaker 1>Specific agent risks listed are collecting too much data lack

432
00:23:49.480 --> 00:23:54.839
<v Speaker 1>of user transparency, control, potential profiling, discrimination, security holes, sharing

433
00:23:54.920 --> 00:23:59.119
<v Speaker 1>data without consent, blurring roles of data, controller processor invading

434
00:23:59.200 --> 00:24:04.599
<v Speaker 1>user autonomy via unsolicited actions. Multi agent systems add more complexity.

435
00:24:04.160 --> 00:24:09.039
<v Speaker 2>Summarized as two risk types one accidental non malicious misuse errors,

436
00:24:09.039 --> 00:24:13.839
<v Speaker 2>bad security, unexpected behavior, two malicious usage insiders selling data,

437
00:24:13.880 --> 00:24:14.680
<v Speaker 2>external attacks.

438
00:24:14.759 --> 00:24:17.720
<v Speaker 1>Do agents actually hold personal data? If not? Less legal

439
00:24:17.799 --> 00:24:18.880
<v Speaker 1>risks likely yes.

440
00:24:18.799 --> 00:24:22.680
<v Speaker 2>Especially delivery, financial habit data in profiles defines personal data

441
00:24:22.680 --> 00:24:25.000
<v Speaker 2>info on identify identifiable natural persons.

442
00:24:25.200 --> 00:24:28.359
<v Speaker 1>Agents aren't data subjects must be natural person, but agent

443
00:24:28.400 --> 00:24:31.160
<v Speaker 1>parameters code could be personal data about the user if linked.

444
00:24:31.319 --> 00:24:33.119
<v Speaker 1>Shopping agents likely store profiles.

445
00:24:33.319 --> 00:24:37.640
<v Speaker 2>Privacy and distributed computing focuses on control who controls agent

446
00:24:37.920 --> 00:24:42.519
<v Speaker 2>data processing location. Agents are means for processing, but determining

447
00:24:42.519 --> 00:24:44.559
<v Speaker 2>the controller is hard. In distributed agent.

448
00:24:44.359 --> 00:24:48.240
<v Speaker 1>Systems, agent autonomy learning challenge. Traditional consent models based on

449
00:24:48.240 --> 00:24:52.599
<v Speaker 1>initial notification. Agents should maybe inform users of processing keep records.

450
00:24:53.039 --> 00:24:56.599
<v Speaker 1>Controllers might want to distance themselves from autonomous agent acts.

451
00:24:56.839 --> 00:25:01.039
<v Speaker 2>Various proposals for online in privacy policies often long hard

452
00:25:01.079 --> 00:25:05.319
<v Speaker 2>for agents. No guarantee of respect seals trusting bb bone

453
00:25:05.359 --> 00:25:09.559
<v Speaker 2>line set codes of conduct, industry codes, FEEDMA, DMA, trusted

454
00:25:09.839 --> 00:25:15.000
<v Speaker 2>third parties, infomediaries, Passport Liberty Alliance, technical solutions, crypto anonymity

455
00:25:15.000 --> 00:25:15.680
<v Speaker 2>tools p THREEP.

456
00:25:15.880 --> 00:25:19.279
<v Speaker 1>P THREEP Platform for Privacy Preferences aims to inform users

457
00:25:19.359 --> 00:25:22.079
<v Speaker 1>via machine readable policies automated handshake, but.

458
00:25:22.119 --> 00:25:26.039
<v Speaker 2>Criticized doesn't enforce compliance. P THREEP compliance doesn't guarantee EU

459
00:25:26.119 --> 00:25:29.240
<v Speaker 2>rule adherence, costly, hard to create accurate policies versus one

460
00:25:29.240 --> 00:25:32.200
<v Speaker 2>point zero maybe two simple Later versions aim higher. Mentions

461
00:25:32.240 --> 00:25:34.880
<v Speaker 2>related work on preference negotiations, DAN, semantic.

462
00:25:34.480 --> 00:25:38.440
<v Speaker 1>Web and IPv six again, embedded digital ID via MC

463
00:25:38.559 --> 00:25:42.920
<v Speaker 1>address potential for tracking despite privacy enhancement.

464
00:25:42.480 --> 00:25:45.759
<v Speaker 2>Work Yes concludes privacy is fundamental for agents in the

465
00:25:45.799 --> 00:25:51.039
<v Speaker 2>research scenario due to automation profiles, data mining needs proactive privacy,

466
00:25:51.079 --> 00:25:55.000
<v Speaker 2>governance for compliance and user trust design agents with privacy

467
00:25:55.000 --> 00:25:55.880
<v Speaker 2>and mind from the start.

468
00:25:56.160 --> 00:25:58.759
<v Speaker 1>Okay, let's wrap up with the conclusions. The book starts

469
00:25:58.759 --> 00:26:01.240
<v Speaker 1>by saying the current way e commerce apps are built

470
00:26:01.400 --> 00:26:03.000
<v Speaker 1>often leads to illegality.

471
00:26:03.200 --> 00:26:06.640
<v Speaker 2>Yeah, because law struggles to keep pace with tech. Legislation

472
00:26:06.680 --> 00:26:09.960
<v Speaker 2>targets existing interactions, finds it hard to anticipate new ones.

473
00:26:10.400 --> 00:26:14.839
<v Speaker 2>Sites delays defining directing activities for Brussels regulation debates on

474
00:26:15.000 --> 00:26:17.200
<v Speaker 2>information society services.

475
00:26:16.920 --> 00:26:21.079
<v Speaker 1>E reluctance on national contract law leaves ambiguity Online tech

476
00:26:21.079 --> 00:26:24.240
<v Speaker 1>moves faster than law to apps seen with privacy directive updates,

477
00:26:24.279 --> 00:26:28.119
<v Speaker 1>ipeer software patent debates. Agents now challenge even recent frameworks

478
00:26:28.160 --> 00:26:31.799
<v Speaker 1>like e signatures, data controller roles. Agent IPR leads.

479
00:26:31.519 --> 00:26:35.279
<v Speaker 2>To retrofitting building tech first. Worrying about law later results

480
00:26:35.279 --> 00:26:38.839
<v Speaker 2>in illegal or partially illegal systems due to mismatch book

481
00:26:38.839 --> 00:26:45.440
<v Speaker 2>revisits tech solutions for trust compliance crypto labels, smart cards, tokens, biometrics.

482
00:26:44.720 --> 00:26:50.279
<v Speaker 1>Watermarks, specific legal process requirements, ISSP identification, e commerce directive

483
00:26:50.400 --> 00:26:54.920
<v Speaker 1>ART five name, address, contact, etc. Agent providers should comply

484
00:26:55.039 --> 00:26:59.440
<v Speaker 1>for confidence invoicing. New VAT directive allows e invoices, maybe

485
00:26:59.480 --> 00:27:02.480
<v Speaker 1>needing digital signatures. National variations exist.

486
00:27:02.759 --> 00:27:05.759
<v Speaker 2>Advocates shifting from low level code to higher level models

487
00:27:05.759 --> 00:27:09.519
<v Speaker 2>for e commerce systems better reflex workflows, easier to integrate

488
00:27:09.599 --> 00:27:12.359
<v Speaker 2>constraints business and legal argues.

489
00:27:12.400 --> 00:27:16.079
<v Speaker 1>Business rules and legal rules are similar. Both restrict process freedom.

490
00:27:16.160 --> 00:27:19.559
<v Speaker 1>Same modeling methods can create consumptual models for legal principles,

491
00:27:19.680 --> 00:27:24.279
<v Speaker 1>privacy contract IPR codifying law structurally.

492
00:27:23.799 --> 00:27:28.519
<v Speaker 2>Proposes standardized model affirm processes allows a generic legal architecture

493
00:27:28.519 --> 00:27:32.000
<v Speaker 2>model for online activities. Apply this legal model to legalize

494
00:27:32.000 --> 00:27:35.279
<v Speaker 2>the business process model automating that leads to more compliant

495
00:27:35.279 --> 00:27:36.599
<v Speaker 2>transactions via agents.

496
00:27:36.799 --> 00:27:38.799
<v Speaker 1>Examples of applying this process.

497
00:27:38.480 --> 00:27:42.519
<v Speaker 2>View agent negotiation mentioned suite deal project using rules rule

498
00:27:42.559 --> 00:27:45.920
<v Speaker 2>mel DEMA plus oil to model contract life cycle exceptions.

499
00:27:46.319 --> 00:27:49.279
<v Speaker 2>Trust and reputation vital between agency users and agent agent

500
00:27:49.720 --> 00:27:54.039
<v Speaker 2>need reliability, security honesty essential for open e commerce, overcoming

501
00:27:54.039 --> 00:27:59.359
<v Speaker 2>resistance to delegation processes for trust authorization, certification, access controls,

502
00:27:59.359 --> 00:28:04.920
<v Speaker 2>security poble reputation mechanisms need to model rights, obligations, permissions, prohibitions,

503
00:28:05.000 --> 00:28:09.000
<v Speaker 2>non fulfillment actions, Reputation systems track trust levels so stable.

504
00:28:09.039 --> 00:28:13.039
<v Speaker 1>Agent e commerce needs common language ontology, norms, rules, technical

505
00:28:13.039 --> 00:28:16.839
<v Speaker 1>and legal standardization happening at language intology level, but pre

506
00:28:16.960 --> 00:28:20.400
<v Speaker 1>modeling at normative legal level. Need common loss protocols for

507
00:28:20.440 --> 00:28:23.079
<v Speaker 1>agent interaction or negotiation rules. P three P is a

508
00:28:23.079 --> 00:28:25.680
<v Speaker 1>partial non agent example but limited legally, though it shows

509
00:28:25.720 --> 00:28:27.640
<v Speaker 1>interaction models that could embed rules.

510
00:28:27.680 --> 00:28:32.240
<v Speaker 2>The ultimate conclusion a process oriented approach integrating legal stress

511
00:28:32.279 --> 00:28:35.720
<v Speaker 2>at design time. Legally, engineering the code is needed for

512
00:28:35.759 --> 00:28:39.599
<v Speaker 2>greater legal certainty and automated contracting reduced risk in Europe

513
00:28:40.119 --> 00:28:44.200
<v Speaker 2>acknowledges some conceptual risks around purely agent deals consent need

514
00:28:44.319 --> 00:28:48.720
<v Speaker 2>judicial legislative clarification, put agents in tech frameworks that already

515
00:28:48.720 --> 00:28:52.960
<v Speaker 2>include a legal dimension. Cites web service frameworks like UDDI, rosette,

516
00:28:53.000 --> 00:28:58.279
<v Speaker 2>net ebXML having pre defined process ontologies, and envisions interoperable

517
00:28:58.279 --> 00:29:02.880
<v Speaker 2>agents offering services via contract referencing shared legal ontologies xml

518
00:29:03.039 --> 00:29:06.839
<v Speaker 2>RDF bound by embedded regulation, learning from legal updates.

519
00:29:07.000 --> 00:29:09.920
<v Speaker 1>It's clear from our deep dive today that designing legally

520
00:29:09.960 --> 00:29:13.480
<v Speaker 1>sound systems in this fast moving digital world, especially with

521
00:29:13.519 --> 00:29:17.880
<v Speaker 1>autonomous agents emerging, is It's complex but absolutely vital.

522
00:29:17.680 --> 00:29:20.559
<v Speaker 2>M HM and as our look at legal programming showed,

523
00:29:21.000 --> 00:29:24.519
<v Speaker 2>taking that proactive approach weaving legal thinking right into the design,

524
00:29:24.720 --> 00:29:26.440
<v Speaker 2>it's not just about avoiding trouble.

525
00:29:26.200 --> 00:29:28.720
<v Speaker 1>No, it's about building the trust needed for these technologies

526
00:29:28.759 --> 00:29:29.559
<v Speaker 1>to really take off.

527
00:29:29.839 --> 00:29:34.400
<v Speaker 2>Exactly understanding this process view the need for legal programming,

528
00:29:34.920 --> 00:29:38.200
<v Speaker 2>it gives you a valuable framework for navigating this intersection

529
00:29:38.240 --> 00:29:39.039
<v Speaker 2>of law and tech.

530
00:29:39.400 --> 00:29:42.519
<v Speaker 1>So the ongoing conversation between legal folks and tech folks

531
00:29:42.799 --> 00:29:46.319
<v Speaker 1>developing solutions that bridge that gap between law and innovation.

532
00:29:46.440 --> 00:29:49.119
<v Speaker 2>It's going to crucial and absolutely crucial for ensuring a

533
00:29:49.160 --> 00:29:53.559
<v Speaker 2>future of digital commerce that's not just innovative, but fundamentally trustworthy.