WEBVTT

00:00:02.000 --> 00:00:06.270
Announcer: This is the AI Law Wiki News Evening Podcast.

00:00:06.350 --> 00:00:11.730
Wadilya Saw: It’s Wednesday, November eleventh, twenty twenty-six and these are the top stories tonight.

00:00:12.390 --> 00:00:17.110
Wadilya Saw: A man says a facial-recognition match sent him to jail for more than two weeks.

00:00:17.670 --> 00:00:21.650
Wadilya Saw: A Kentucky judge lets his lawsuit against the investigating detective proceed.

00:00:22.450 --> 00:00:31.770
Wadilya Saw: And Europe proposes new controls for AI agents that spend your money, including a separate approval when commercial interests could influence the purchase.

00:00:32.490 --> 00:00:33.270
Wadilya Saw: I’m Wadilya Saw.

00:00:33.290 --> 00:00:34.750
Max Token: And I’m Max Token.

00:00:35.450 --> 00:00:39.430
Max Token: We’re AI-generated voices reading an AI-generated script.

00:00:41.630 --> 00:00:51.250
Wadilya Saw: Alain Vizcay Osorio says an AI facial match wrongly linked him to a suspected check fraud in Kentucky and sent him to jail for more than two weeks.

00:00:52.050 --> 00:00:59.490
Wadilya Saw: On Monday, November ninth, a federal judge allowed his civil rights lawsuit against the investigating detective to move forward.

00:01:00.030 --> 00:01:05.290
Wadilya Saw: The case turns on how a computer’s suggested match became the basis for an arrest warrant.

00:01:06.170 --> 00:01:17.830
Wadilya Saw: According to Osorio’s complaint, Florence detective Ryan Kilmer took a screenshot from a bank’s security camera and used facial-recognition technology to find a booking photo and a Facebook image.

00:01:18.370 --> 00:01:26.760
Wadilya Saw: Osorio says Kilmer relied entirely on the AI identification, despite visible differences between his appearance and the suspect’s.

00:01:27.720 --> 00:01:33.680
Wadilya Saw: Osorio also alleges that Kilmer deliberately or recklessly misled the court issuing the warrant.

00:01:34.420 --> 00:01:44.580
Wadilya Saw: The detective described investigative tools and his own comparison of the pictures, without explaining that the identification rested on an AI result he hadn’t corroborated.

00:01:45.110 --> 00:01:48.210
Wadilya Saw: The suspected check fraud happened in August twenty twenty-four.

00:01:49.170 --> 00:01:51.790
Wadilya Saw: Osorio was arrested in January twenty twenty-five.

00:01:52.990 --> 00:02:00.450
Wadilya Saw: According to the court’s account, the charges were dismissed that May after he proved he’d been working in Nebraska when the check was presented.

00:02:00.990 --> 00:02:04.710
Wadilya Saw: He brought this damages suit against Kilmer in January this year.

00:02:05.450 --> 00:02:07.670
Wadilya Saw: At this early stage, Judge S.

00:02:07.990 --> 00:02:13.070
Wadilya Saw: Chad Meredith accepts the complaint’s factual allegations to decide whether the case can proceed.

00:02:13.770 --> 00:02:23.570
Wadilya Saw: Kilmer argues that the warrant contained enough information to justify the arrest, and that failing to double-check a match isn’t enough to establish a constitutional violation.

00:02:24.270 --> 00:02:30.790
Wadilya Saw: He also says no previous decision specifically prohibited relying solely on AI facial recognition.

00:02:31.630 --> 00:02:40.430
Wadilya Saw: Meredith’s answer is that established arrest standards still apply: officers need a reasonable basis to believe the person committed the offense.

00:02:41.290 --> 00:02:47.350
Wadilya Saw: On Osorio’s account, the misleading statements supplied the essential link between him and the bank suspect.

00:02:48.050 --> 00:02:52.150
Wadilya Saw: A computer lead required independent judgment and supporting investigation.

00:02:52.170 --> 00:03:01.150
Wadilya Saw: The judge also declined to grant qualified immunity, the protection for officers whose conduct doesn’t violate clearly established rights.

00:03:01.970 --> 00:03:07.980
Wadilya Saw: He found the alleged conduct sufficiently obviously unlawful without an AI-specific precedent.

00:03:08.520 --> 00:03:14.780
Wadilya Saw: Osorio’s civil case can now move forward to examine how the investigation led to his arrest and detention.

00:03:15.510 --> 00:03:23.050
Max Token: Job applicants aged forty and older say Workday’s AI-assisted recruiting tools put them at a disadvantage.

00:03:23.990 --> 00:03:26.730
Max Token: Workday supplies hiring software to employers.

00:03:27.510 --> 00:03:34.310
Max Token: More than fifteen thousand applicants have joined a civil age-discrimination case challenging that technology together.

00:03:35.010 --> 00:03:38.170
Max Token: Now Workday wants the court to break up that group.

00:03:38.970 --> 00:03:44.230
Max Token: Sheila Callaham reports for Forbes that the company filed its request yesterday, November tenth.

00:03:44.250 --> 00:03:47.470
Max Token: The dispute is about what connects these applicants.

00:03:48.210 --> 00:03:54.150
Max Token: They sought different jobs at different companies, but they’re bringing claims against the same software provider.

00:03:54.690 --> 00:04:02.810
Max Token: Workday argues that its customers configure and use the tools differently, set their own job requirements and make their own hiring decisions.

00:04:03.470 --> 00:04:09.190
Max Token: It also says the applicants haven’t shown that its AI evaluated most of the applications involved.

00:04:09.210 --> 00:04:16.370
Max Token: In May twenty twenty-five, a federal judge gave preliminary permission for eligible applicants to join the case.

00:04:16.890 --> 00:04:20.590
Max Token: Workday’s motion asks the judge to end that collective process.

00:04:21.110 --> 00:04:23.730
Max Token: Individual claims could still be pursued separately.

00:04:24.390 --> 00:04:26.290
Max Token: The judge hasn’t ruled on this request.

00:04:26.870 --> 00:04:36.170
Max Token: The question is whether the applicants can identify a common discriminatory practice, or whether their claims depend on each employer’s individual circumstances.

00:04:37.030 --> 00:04:40.890
Wadilya Saw: An AI shopping agent can do more than recommend a product.

00:04:41.550 --> 00:04:45.110
Wadilya Saw: It can enter a contract for you, committing your money to a purchase.

00:04:45.830 --> 00:04:54.790
Wadilya Saw: Today, November eleventh, the European Commission proposed rules giving consumers control over when that software must stop and ask permission.

00:04:55.550 --> 00:04:58.610
Wadilya Saw: The rules are part of the proposed Digital Fairness Act.

00:04:59.190 --> 00:05:05.130
Wadilya Saw: They would address both how much authority a shopper delegates and whose commercial interests the agent might serve.

00:05:05.150 --> 00:05:09.510
Wadilya Saw: On first use, consumers would have to make an explicit choice.

00:05:10.210 --> 00:05:20.830
Wadilya Saw: They could require approval of every contract, require approval only for certain contracts under criteria they set, or authorize purchases without individual confirmation.

00:05:21.810 --> 00:05:25.490
Wadilya Saw: Those criteria could include a maximum price and a time limit.

00:05:26.070 --> 00:05:29.310
Wadilya Saw: Consumers could change their choice and settings at any time.

00:05:29.930 --> 00:05:35.530
Wadilya Saw: Imagine authorizing an agent to make purchases below a price limit without checking each one with you.

00:05:35.950 --> 00:05:37.550
Wadilya Saw: That could save repeated approvals.

00:05:38.210 --> 00:05:42.750
Wadilya Saw: But suppose the agent’s provider receives payment for proposing a particular purchase.

00:05:43.350 --> 00:05:52.690
Wadilya Saw: Under the proposal, the system would have to disclose that payment and ask you to approve the contract separately, even though you’d generally authorized automatic buying.

00:05:52.710 --> 00:05:56.910
Wadilya Saw: The same requirement would apply to a conflict of interest with the merchant.

00:05:57.570 --> 00:06:04.830
Wadilya Saw: The concern is that a recommendation could serve the provider’s commercial relationship while the software is acting on the buyer’s behalf.

00:06:05.290 --> 00:06:11.870
Wadilya Saw: Disclosure would explain that relationship; separate confirmation would give the buyer the decision on that transaction.

00:06:12.370 --> 00:06:18.810
Wadilya Saw: Merchants also couldn’t require consumers to use automated systems to access or complete online transactions.

00:06:18.830 --> 00:06:23.230
Wadilya Saw: These particular contracting rules would exclude financial services.

00:06:23.910 --> 00:06:29.830
Wadilya Saw: The wider package would also tackle online practices such as obstacles to cancelling subscriptions.

00:06:30.590 --> 00:06:38.150
Wadilya Saw: The proposal now needs agreement from the European Parliament and the Council before these controls could become EU consumer law.

00:06:38.890 --> 00:06:45.870
Max Token: In Britain, a fake takeover bid used AI-forged paperwork to make a fictitious buyer look credible.

00:06:45.890 --> 00:06:52.590
Max Token: The intended payoff depended on investors believing the bid and driving up the company’s share price.

00:06:53.270 --> 00:07:04.990
Max Token: The Financial Conduct Authority says Christopher Woolcott bought a stake in oil and gas company Touchstone Exploration, then staged a takeover offer worth a hundred and fifty-four million pounds.

00:07:05.910 --> 00:07:14.090
Max Token: He used ChatGPT to forge documents, invented identities and met the company’s chief executive under a false name.

00:07:14.110 --> 00:07:21.930
Max Token: The regulator says Woolcott could have gained as much as seventy thousand pounds by selling his shares if the plan had worked.

00:07:22.570 --> 00:07:24.790
Max Token: Touchstone determined that the bid was false.

00:07:25.410 --> 00:07:31.710
Max Token: Woolcott pleaded guilty on September tenth to fraud by false representation and three forgery offenses.

00:07:32.870 --> 00:07:41.630
Max Token: Yesterday, November tenth, Southwark Crown Court gave him a two-year prison sentence, suspended for two years, and ordered two hundred hours of unpaid work.

00:07:41.650 --> 00:07:46.270
Max Token: The suspended sentence means he isn’t being sent immediately to prison.

00:07:47.210 --> 00:07:57.290
Wadilya Saw: Brazil has approved education guidelines that would bar schools and universities from punishing students solely because an automated detector labels their work AI-generated.

00:07:58.690 --> 00:08:02.570
Wadilya Saw: Those detectors assess whether content was produced by artificial intelligence.

00:08:03.630 --> 00:08:09.630
Wadilya Saw: Under the guidelines, that classification alone couldn’t justify an academic or disciplinary sanction.

00:08:09.650 --> 00:08:16.830
Wadilya Saw: The distinction matters when a suspicion about authorship could affect a student’s grade or ability to remain in school.

00:08:17.390 --> 00:08:27.650
Wadilya Saw: Yesterday, November tenth, Brazil’s Education Ministry approved the National Education Council’s opinion covering public and private education at every level.

00:08:28.090 --> 00:08:35.870
Wadilya Saw: The guidelines keep people responsible for consequential decisions, including assessment, progression, certification and sanctions.

00:08:35.890 --> 00:08:43.490
Wadilya Saw: AI could support teaching and administration, while educators would retain responsibility for educational judgments.

00:08:44.030 --> 00:08:49.150
Wadilya Saw: According to the government’s announcement, the prohibitions take effect when the resolution is published.

00:08:49.750 --> 00:08:58.690
Wadilya Saw: That publication also starts a twelve-month period for education systems and institutions to adapt their policies, practices and contracts.

00:09:00.450 --> 00:09:01.950
Max Token: That’s it for tonight’s podcast.

00:09:02.670 --> 00:09:05.150
Max Token: Our podcast is written and produced by Codex.

00:09:05.890 --> 00:09:08.410
Max Token: Our voices are provided by ElevenLabs.

00:09:08.870 --> 00:09:10.870
Max Token: Our theme music was composed by Gemini.

00:09:11.470 --> 00:09:12.330
Max Token: I’m Max Token.

00:09:12.690 --> 00:09:14.050
Wadilya Saw: And I’m Wadilya Saw.

00:09:14.250 --> 00:09:15.230
Max Token: Thank you for listening.

00:09:15.590 --> 00:09:18.670
Max Token: We’ll see you next time on the AI Law Wiki News Podcast.
