Announcer: This is the AI Law Wiki News Evening Podcast.
Wadilya Saw: Good evening.
Wadilya Saw: I'm Wadilya Saw.
Max Token: And I'm Max Token.
Max Token: It's Monday, October fifth, twenty twenty-six.
Wadilya Saw: Tonight: OpenAI begins a text-watermarking rollout in response to European transparency rules.
Wadilya Saw: Norway prepares restrictions on AI glasses.
Wadilya Saw: And Washington names the leader of a new federal AI task force.
Max Token: This report and our voices were created with generative AI.
Wadilya Saw: OpenAI says customers using its programming interface can now choose text watermarking for selected models worldwide.
Wadilya Saw: The option starts today, October fifth, and remains off by default.
Wadilya Saw: For eligible ChatGPT and Codex text in the European Union, the company plans to add invisible watermarks over the coming weeks.
Wadilya Saw: It is opening applications for a detector, initially for approved researchers and expert organizations.
Max Token: The legal background is Article fifty of the EU AI Act.
Max Token: It requires providers of systems that generate synthetic content to mark outputs in a machine-readable form so they can be detected as artificial.
Max Token: The text covers audio, images, video, and writing.
Max Token: It asks for technical solutions that are effective and reliable, taking account of what is technically feasible.
Max Token: Standard editing assistance has a specified exception.
Max Token: That's why the exact function of a tool matters when applying the rule.
Wadilya Saw: OpenAI's approach puts a statistical signal in word choices.
Wadilya Saw: Its detector searches for that pattern.
Wadilya Saw: The company acknowledges both missed signals and false positives.
Wadilya Saw: Short or edited passages are harder to assess.
Wadilya Saw: In one company evaluation, replacing ten percent of the words with synonyms reduced detection from roughly ninety-two percent to sixty-six percent.
Wadilya Saw: OpenAI says those limitations explain its restricted initial detector access.
Max Token: The Act also separates provider marking from disclosure by people using AI.
Max Token: For some public-interest text, human review, editorial control, and responsibility for publication affect the disclosure rule.
Max Token: A provider's machine-readable mark and a publisher's notice answer different questions.
Max Token: One concerns identifying the output's origin; the other concerns informing the audience in a particular use.
Wadilya Saw: OpenAI says a detector result gives limited evidence about its systems' involvement.
Wadilya Saw: Ownership, responsibility, and factual accuracy require separate assessment.
Wadilya Saw: The next steps are the EU rollout and further evaluation of the detector.
Wadilya Saw: For listeners, the useful distinction is between a technical provenance signal and the judgment needed to assess the writing itself.
Max Token: Norway's government says it will bring Parliament a bill for temporary restrictions on AI glasses as soon as possible.
Max Token: Today's announcement centers on people being recorded without knowing it.
Max Token: Places under consideration include parks, beaches, schools, childcare settings, and healthcare facilities.
Max Token: Changing rooms and showers receive particular attention.
Max Token: The government will define the covered technology during drafting, including how cameras, sound recording, and AI functions fit together.
Wadilya Saw: The government is also considering exceptions for vulnerable groups and limited uses that benefit society.
Wadilya Saw: Private use would remain possible.
Wadilya Saw: Alongside the temporary proposal, an expert group would examine permanent national rules for wearable technology.
Wadilya Saw: Those choices matter: the device definition, the places covered, and the exceptions determine whose use would actually be restricted.
Max Token: The next step is published legislation for the Storting, Norway's Parliament.
Max Token: The announcement describes a proposed ban, with drafting still ahead.
Max Token: It gives the privacy concern and the intended route to legislation; the bill will supply the operative terms.
Max Token: We'll follow that text and Parliament's response.
Wadilya Saw: In Washington, President Donald Trump has named Director of National Intelligence Jay Clayton to lead a new federal AI task force.
Wadilya Saw: The announcement came yesterday, October fourth.
Wadilya Saw: Trump calls the group the Super Intelligence Force.
Wadilya Saw: According to the Associated Press, it will coordinate federal efforts and reach out to consumers, public-interest groups, religious organizations, critical-infrastructure providers, and AI companies.
Max Token: Other members include Federal Trade Commission chair Andrew Ferguson, Pentagon technology chief Emil Michael, and Office of Personnel Management director Scott Kupor.
Max Token: The group will report to Trump and his chief of staff, Susie Wiles.
Max Token: That membership puts consumer protection, defense technology, and federal personnel administration in the same coordinating group.
Wadilya Saw: For now, the appointment and announced outreach are the concrete developments.
Wadilya Saw: The practical questions for follow-up are what work the group undertakes, what recommendations it produces, and how agencies act on them.
Wadilya Saw: Today's news is the organization of that effort, announced on Sunday and first covered in the wiki's Monday edition.
Max Token: The UK-based Internet Watch Foundation has released new findings on AI-generated child sexual abuse images.
Max Token: Its analysts assessed six thousand three hundred and ten images meeting the legal definition of child sexual abuse between January and June this year.
Max Token: That compares with four thousand five hundred and twelve in all of twenty twenty-five.
Max Token: These are the foundation's assessed images, with different observation periods, rather than a count of all such material online.
Wadilya Saw: The foundation is urging EU lawmakers negotiating the proposed Child Sexual Abuse Regulation to include tools for detecting previously unseen material.
Wadilya Saw: Its explanation is about digital fingerprints, often called hashes.
Wadilya Saw: Once an image has been identified, a fingerprint can help services recognize known copies.
Wadilya Saw: A newly created image must first be identified before it has that known fingerprint.
Max Token: The foundation argues that a framework confined to known material would leave a gap as AI generates new images.
Max Token: That is its position on the pending proposal.
Max Token: The next issue is how EU negotiators design the legal framework.
Max Token: The figures describe the foundation's work in the first half of this year; the legislative call was published today.
Wadilya Saw: Malaysia's digital minister, Gobind Singh Deo, says the government plans to table its AI Governance Bill in Parliament early next year.
Wadilya Saw: He spoke during an October fifth question session in the lower house, the Dewan Rakyat.
Wadilya Saw: The Edge Malaysia reports that the government is drafting controls according to the potential harm of the AI system, across sectors and alongside existing sectoral laws.
Max Token: The minister also described safety testing, monitoring, human oversight, and response mechanisms as systems become more powerful.
Max Token: He said the Malaysia AI Safety Institute would work on safety benchmarks, risk assessments, and incident reporting.
Max Token: Those are the government's stated aims; the published bill will establish the exact proposed duties.
Wadilya Saw: Early twenty twenty-seven is the announced tabling timetable.
Wadilya Saw: The bill remains in preparation.
Wadilya Saw: For businesses and the public, the next source to examine is the introduced text: who it covers, what oversight it creates, and how obligations are assigned.
Wadilya Saw: Those details will show how the minister's system-based risk approach translates into legislation.
Max Token: That’s our evening report.
Max Token: Good night.