WEBVTT

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

00:00:06.320 --> 00:00:11.740
Wadilya Saw: It's Friday, October thirtieth, twenty twenty-six and these are the top stories tonight.

00:00:12.340 --> 00:00:16.320
Wadilya Saw: Britain's competition regulator investigates Meta's content terms.

00:00:16.820 --> 00:00:20.740
Wadilya Saw: Major AI providers announce a shared training preference for websites.

00:00:21.780 --> 00:00:27.840
Wadilya Saw: And copyright advocates offer sharply different answers about who should control the use of human work.

00:00:28.300 --> 00:00:29.420
Wadilya Saw: I'm Wadilya Saw.

00:00:29.700 --> 00:00:30.660
Max Token: And I'm Max Token.

00:00:30.680 --> 00:00:35.400
Max Token: We're AI-generated voices reading an AI-generated script.

00:00:37.600 --> 00:00:43.560
Wadilya Saw: Britain's Competition and Markets Authority opened an investigation today into Meta's commercial terms.

00:00:44.280 --> 00:00:52.980
Wadilya Saw: The question is whether businesses that depend on Meta to reach their audiences are being required to accept unfair conditions on the use of their content.

00:00:53.000 --> 00:01:08.040
Max Token: The regulator says the terms could let Meta collect material from publishers' and other businesses' websites, then use it in its own products, including AI features, without giving those businesses a choice about that use.

00:01:08.880 --> 00:01:17.640
Max Token: The investigation will first examine whether Meta holds a dominant position in a relevant British market, then whether the terms abuse that position.

00:01:17.660 --> 00:01:22.440
Wadilya Saw: That is a competition-law inquiry, with evidence still to be gathered.

00:01:23.160 --> 00:01:26.360
Wadilya Saw: The authority has made no decision that Meta broke the law.

00:01:26.980 --> 00:01:28.320
Wadilya Saw: And the rollout has changed.

00:01:29.000 --> 00:01:39.040
Wadilya Saw: Meta originally planned the revision for today, delayed it to November twentieth, then announced a pause on October twenty-eighth pending consultation with industry groups.

00:01:39.540 --> 00:01:43.780
Max Token: So November twentieth is no longer a firm launch date in this account.

00:01:44.500 --> 00:01:54.900
Max Token: The practical issue is the bargaining power behind a platform's terms: whether a business can meaningfully refuse a new use of its work while continuing to reach customers.

00:01:55.800 --> 00:02:00.220
Max Token: The regulator will now seek evidence from Meta and affected third parties.

00:02:00.620 --> 00:02:04.000
Wadilya Saw: Publishers also received a different kind of announcement today.

00:02:04.640 --> 00:02:13.240
Wadilya Saw: OpenAI, Google and Microsoft say they will support a shared way for websites to express preferences about generative AI training.

00:02:13.900 --> 00:02:20.180
Wadilya Saw: Website operators would put the signal in robots dot text, the file crawlers consult for instructions.

00:02:21.000 --> 00:02:26.540
Max Token: The idea is to reduce the need to maintain a separate training rule for every provider.

00:02:27.140 --> 00:02:30.360
Max Token: But today's announcements are commitments to future implementation.

00:02:31.400 --> 00:02:33.360
Max Token: OpenAI says the coming months.

00:02:34.160 --> 00:02:42.820
Max Token: Google aims for twenty twenty-seven and says existing explicit Google Extended choices will continue to work alongside the new signal.

00:02:43.320 --> 00:02:47.400
Wadilya Saw: Microsoft supplies more specific dates and a second distinction.

00:02:48.040 --> 00:02:50.640
Wadilya Saw: Training changes a model's learned parameters.

00:02:51.380 --> 00:02:55.560
Wadilya Saw: Grounding uses content to help inform an answer when it is generated.

00:02:56.200 --> 00:03:04.660
Wadilya Saw: Microsoft's plan lets publishers express independent preferences for those two uses, while remaining eligible for ordinary Bing search results.

00:03:04.680 --> 00:03:12.180
Max Token: Microsoft says it will begin honoring the new preferences on February first, twenty twenty-seven, as it crawls.

00:03:12.560 --> 00:03:20.100
Max Token: On March first, its older No Archive and No Cache instructions will stop expressing training or grounding preferences.

00:03:20.800 --> 00:03:28.940
Max Token: Existing users need to review the transition steps; publishing a new page instruction does not guarantee it has already been discovered.

00:03:28.960 --> 00:03:34.800
Wadilya Saw: These are participating companies' usage controls, based on standards work still in progress.

00:03:35.520 --> 00:03:41.940
Wadilya Saw: They do not establish a universal opt-out, a payment agreement, or a court's answer to the copyright question.

00:03:42.560 --> 00:03:47.720
Wadilya Saw: They do show why training and answer generation increasingly need separate explanations.

00:03:48.200 --> 00:03:51.600
Max Token: That same distinction runs through today's copyright coverage.

00:03:52.240 --> 00:04:03.500
Max Token: The Wikimedia Foundation says transformative model training can qualify as fair use, while copied answers that substitute for source material should respect licensing requirements.

00:04:04.320 --> 00:04:08.620
Max Token: For Wikipedia content, that includes attribution and share-alike terms.

00:04:09.280 --> 00:04:17.460
Wadilya Saw: Wikimedia explained its position today in the consolidated OpenAI and Microsoft copyright litigation in federal court in New York.

00:04:18.180 --> 00:04:24.860
Wadilya Saw: Its October twenty-ninth letter asks permission to submit a proposed friend-of-the-court brief supporting neither side.

00:04:24.880 --> 00:04:31.200
Wadilya Saw: The reviewed record does not establish an order accepting it, or a decision on fair use.

00:04:31.700 --> 00:04:34.420
Max Token: Publishing organizations take a different position.

00:04:35.140 --> 00:04:40.740
Max Token: The Association of American Publishers and three partner groups announced their submission yesterday.

00:04:41.660 --> 00:04:52.240
Max Token: They support the plaintiffs' request for judgment, arguing that licensing markets and the sustainability of human authorship weigh against the defendants' fair-use position.

00:04:52.850 --> 00:04:57.840
Wadilya Saw: Authors Alliance announced its own position today, after requesting permission yesterday.

00:04:58.640 --> 00:05:02.560
Wadilya Saw: It supports the defendants on a narrower issue: market dilution.

00:05:03.300 --> 00:05:11.900
Wadilya Saw: The group argues that competition from newly generated works should not count as copyright injury when those works do not copy protected expression.

00:05:12.440 --> 00:05:18.880
Wadilya Saw: It takes no position on how training material was acquired or whether particular outputs infringe.

00:05:18.900 --> 00:05:21.920
Max Token: Those are competing arguments for the judge to assess.

00:05:22.640 --> 00:05:29.380
Max Token: The disagreement concerns both the act of training and what kinds of economic harm copyright recognizes.

00:05:30.120 --> 00:05:32.580
Max Token: None of these announcements settles the case.

00:05:33.060 --> 00:05:43.520
Wadilya Saw: In India, the telecom regulator released draft guidelines today for licensed or authorized entities using, or planning to use, AI in their networks and services.

00:05:44.380 --> 00:05:46.600
Wadilya Saw: Comments are due November twenty-seventh.

00:05:46.620 --> 00:05:56.620
Max Token: The proposal would require a board-approved internal AI policy, risk classifications, an inventory of systems and checks before deployment.

00:05:57.160 --> 00:06:05.840
Max Token: Operators would assess suppliers, include accountability provisions in contracts and provide human oversight proportionate to the risks.

00:06:06.760 --> 00:06:10.480
Max Token: A committee headed by a senior executive would oversee implementation.

00:06:11.320 --> 00:06:16.940
Wadilya Saw: Responsibility would stay with the telecom entity even when it buys or outsources the AI.

00:06:17.680 --> 00:06:23.920
Wadilya Saw: Critical and high-severity incidents would have to be reported to the regulator in a prescribed format and timeframe.

00:06:24.900 --> 00:06:28.740
Wadilya Saw: The reviewed draft does not specify a number of hours for those reports.

00:06:29.300 --> 00:06:31.580
Max Token: This remains a consultation draft.

00:06:32.200 --> 00:06:42.880
Max Token: The immediate next step is feedback on how these proposed safeguards should work across telecom operations, rather than compliance with a newly effective set of duties.

00:06:43.760 --> 00:06:51.500
Wadilya Saw: The U.S. Securities and Exchange Commission announced civil charges yesterday against BG Wealth Sharing and DSJ Exchange.

00:06:52.440 --> 00:07:00.940
Wadilya Saw: It alleges that a purported crypto-investment program raised more than one hundred thirty million dollars from tens of thousands of retail investors.

00:07:00.960 --> 00:07:09.900
Max Token: According to the agency, the program promised extraordinary, risk-free returns using purported automated trading signals.

00:07:10.600 --> 00:07:20.500
Max Token: The SEC says the defendants used false regulatory filings and fabricated certificates bearing the agency's seal to suggest official approval.

00:07:21.220 --> 00:07:26.080
Max Token: The alleged scheme ran from October twenty twenty-five through June twenty twenty-six.

00:07:26.100 --> 00:07:29.540
Wadilya Saw: The case is in federal court in Northern California.

00:07:30.180 --> 00:07:36.440
Wadilya Saw: The SEC seeks permanent injunctions, repayment of ill-gotten gains with interest, and civil penalties.

00:07:37.380 --> 00:07:40.780
Wadilya Saw: Those remedies are requests in a complaint, not a judgment.

00:07:41.220 --> 00:07:44.720
Max Token: The AI-related investment pitch is part of the allegations.

00:07:45.600 --> 00:07:56.160
Max Token: Today's account does not establish that a working AI trading system existed, or that filing a form with the SEC gave the investment program regulatory approval.

00:07:56.680 --> 00:07:59.200
Wadilya Saw: A separate data-use case reaches a settlement.

00:07:59.700 --> 00:08:10.020
Wadilya Saw: Oregon's attorney general announced yesterday that General Motors will pay two hundred twenty million dollars in a resolution led by a coalition of thirty-six attorneys general.

00:08:10.520 --> 00:08:18.120
Max Token: The coalition alleged that GM sold identifiable driving and location data to brokers, who sold it to insurers.

00:08:18.800 --> 00:08:23.720
Max Token: According to Oregon, insurers sometimes used that information to change premiums.

00:08:24.620 --> 00:08:32.120
Max Token: The account concerns GM's Smart Driver program and enrollment practices dating from twenty sixteen through twenty twenty-four.

00:08:32.140 --> 00:08:40.480
Wadilya Saw: Oregon says the settlement requires affirmative, express consent before collecting or sharing driving and location data.

00:08:41.340 --> 00:08:49.980
Wadilya Saw: It also requires limits on collection, a privacy program, internal and external assessments, and dealership training and oversight.

00:08:50.360 --> 00:08:56.080
Max Token: For listeners following automated decisions, the connection is the data that can feed those decisions.

00:08:56.540 --> 00:09:00.340
Max Token: This settlement addresses collection, sharing and consent.

00:09:01.020 --> 00:09:04.820
Max Token: It does not announce a general ban on AI insurance tools.

00:09:05.300 --> 00:09:14.260
Wadilya Saw: Finally, Britain's Department for Work and Pensions published a transparency record today for Work Hub, an AI employment-support tool.

00:09:14.880 --> 00:09:24.400
Wadilya Saw: The department describes a voluntary private beta available to several thousand users, offering help with CVs, skills, jobs and training.

00:09:24.420 --> 00:09:30.840
Max Token: The record says the tool does not decide benefit entitlement or automate statutory decisions.

00:09:31.540 --> 00:09:36.480
Max Token: It also says individual interactions do not receive mandatory human review.

00:09:37.120 --> 00:09:41.580
Max Token: Testing, evaluation and monitoring provide the stated oversight.

00:09:41.980 --> 00:09:50.900
Wadilya Saw: The department reports privacy and equality assessments completed in May, along with demographic bias testing and safeguards for personal information.

00:09:51.660 --> 00:09:54.280
Wadilya Saw: Those are the department's descriptions of its controls.

00:09:55.180 --> 00:10:00.560
Wadilya Saw: The publication is not an independent audit or an announcement of a nationwide launch.

00:10:00.760 --> 00:10:09.620
Max Token: It gives users a more concrete picture of where AI advice sits in the service, and where human review does and does not occur.

00:10:10.460 --> 00:10:13.340
Max Token: Links to all of tonight's sources are in the show notes.

00:10:14.990 --> 00:10:16.490
Max Token: That's it for tonight's podcast.

00:10:17.170 --> 00:10:19.630
Max Token: Our podcast is written and produced by Codex.

00:10:20.330 --> 00:10:22.790
Max Token: Our voices are provided by ElevenLabs.

00:10:23.370 --> 00:10:25.470
Max Token: Our theme music was composed by Gemini.

00:10:26.270 --> 00:10:27.010
Max Token: I'm Max Token.

00:10:27.350 --> 00:10:28.670
Wadilya Saw: And I'm Wadilya Saw.

00:10:28.930 --> 00:10:29.890
Max Token: Thank you for listening.

00:10:30.310 --> 00:10:33.470
Max Token: We'll see you next time on the AI Law Wiki News Podcast.
