News
Latest AI Developments
News and updates from the world of AI, curated and hosted here.
AI Washing: Korea Drafts a Guideline on AI Advertising
AI washing is now on Korea's regulatory agenda. On 22 October 2026 the Korea Fair Trade Commission published a draft guideline on AI advertising, open for comment until 11 November. It sets a test for when a product may be called "AI", gives 16 examples, from a heated mat with only a timer to an "AI lawyer" built on a general chatbot, and adds an eight-point checklist. The article compares it with Türkiye's new AI advertising rules and the EU's Unfair Commercial Practices Directive.
Read →
Judge Used AI: 5th Circuit Vacates Injunction and Reassigns Case
A federal judge used AI to prepare a court order, and on 22 October 2026 the US Fifth Circuit vacated his injunction and reassigned the case. Judge Henry Wingate's orders named non-parties as plaintiffs, relied on declarations missing from the record and cited a case that does not exist. What Jackson Federation of Teachers v. Fitch decides and leaves open, and what Turkish, Korean and European rules expect of judges and lawyers who use AI.
Read →
Digital Omnibus GDPR Changes: The Council's Position Explained
The Digital Omnibus GDPR changes moved a step closer on 21 October 2026, when EU ambassadors agreed the Council's negotiating position. The mandate is not law yet. It would keep today's definition of personal data, add a rule on pseudonymised data, confirm legitimate interest as a basis for AI but drop the unconditional right to object, move breach notices to 96 hours and rewrite the cookie rule. What it says, who objects, and how it compares with KVKK and Korea's PIPA.
Read →
SCHUFA BGH Ruling: When a Credit Score Is an Automated Decision
The SCHUFA BGH ruling of 21 October 2026 decides when a credit score is an automated decision under Article 22 GDPR: only where a company's contract decision depends decisively on it. Germany's Federal Court of Justice also gave consumers a presumption: a score sent, then an adverse decision soon after, is enough. What the four judgments say, how they apply the CJEU's SCHUFA case, what changes on 20 November 2026, and how Türkiye's Findeks score and Korea's credit rules compare.
Read →
Machinery Regulation and AI: Türkiye Adopts EU 2023/1230 Rules
Türkiye's new Machinery Regulation, published in the Resmî Gazete on 16 October 2026, brings EU Regulation 2023/1230 into Turkish law and puts machine learning under product safety rules. From 20 January 2027, safety components with self-evolving behaviour need a notified body, and control systems must limit and log what machines learn. What changes for manufacturers, importers and exporters, and how the rules fit with the EU AI Act after the July 2026 omnibus.
Read →
EU AI Act Enforcement Begins: AI Office Questions Frontier Labs
EU AI Act enforcement against the makers of the most powerful AI models has begun. On 14 October 2026 in Dublin, Commission Executive Vice-President Henna Virkkunen said the AI Office had sent its first Requests for Information to frontier developers on security, external evaluation and monitoring, after a summer of reported loss-of-control incidents. The Commission named no companies; Euronews reports OpenAI and Anthropic. Whether the Act reaches models still in testing is already disputed.
Read →
States Suing OpenAI: Eight AGs Allege ChatGPT Misled Parents
Eight US states are suing OpenAI, alleging that ChatGPT was sold to parents as safe for teenagers when it was not. Filed on 14 October 2026 in separate state courts, the suits by Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia rely mainly on consumer-protection law. They seek per-violation penalties, age assurance, parental controls and limits on frontier-model testing. What the complaints say, how OpenAI has responded, and how Korean and Turkish law approach the same problem.
Read →
Law No. 7596: How Investors in Liquidated Turkish Funds Are Paid
On 9 October 2026 AK Parti members of parliament submitted Bill No. 2/3939 to the Grand National Assembly; the General Assembly adopted it on 22 October 2026 as Law No. 7596, published in the Official Gazette on 27 October 2026. It sets out how investors in the funds whose liquidation the Capital Markets Board ordered on 17 September would be paid. Where cash falls short, distribution would follow an inflation-indexed "net investment amount" rather than unit value, with interim payments of up to TRY 1 million per investor in each fund and the option of paying smaller investors first. The bill also creates a ring-fenced recovery account at TMSF, bars payment to persons the Board determines took part in market abuse involving the funds, and takes effect from 17 September. We go through it article by article, work through hypothetical figures and set out the questions it leaves open.
Read →
Italy AI Law: Decree 179/2026 Sets Fines, Voids Algorithm Firings
Italy's AI law gets its enforcement decree. Legislative Decree 179/2026, published on 8 October 2026 and in force from 23 October, names ACN, AgID and the financial supervisors as EU AI Act authorities and sets fines of up to €35 million or 7% of turnover. Its Article 41 makes dismissals decided solely by automated means void, even with the worker's consent. What the decree says, how it compares with KVKK Article 11(1)(g) and Korea's PIPA Article 37-2, and what employers with staff in Italy should check.
Read →
Türkiye Appoints a Head for Its Public AI Directorate
On 1 October 2026 Türkiye's Official Gazette published the decision assigning a Galatasaray University faculty member to head the Public AI Directorate-General. The unit sits inside the Cybersecurity Presidency; a December 2025 presidential decree created it and gave the Presidency duties to lead public-sector AI, set data-governance standards and grant conformity on data quality. On 7 October, according to press reports, the Minister of Industry and Technology described the rest of the AI Action Plan's governance: a National AI Ethics Board without binding powers, pre-deployment impact assessment for public-sector AI in health, education and employment, and regulatory sandboxes. One institution now exists in law and has a head. The others are still plans, and for companies the first rules are likely to arrive through public procurement.
Read →
California SB 574: AI Rules for Lawyers and Arbitrators
On 30 September 2026 California enacted SB 574, which turns court warnings and soft-law guidance on generative AI into statute. From 1 January 2027 attorneys may not delegate the practice of law to AI. They must keep confidential and personal data out of systems that others can access, verify AI output and disclose its use to courts. No filing may contain a citation the responsible attorney has not personally verified. Arbitrators may not delegate any part of their decision-making to AI. We set out what the text says, where it came from, and how it compares with Türkiye and Korea.
Read →
California's Workplace AI Laws: No Firing by Algorithm Alone
On 30 September 2026 California enacted four workplace AI statutes. SB 947 bars employers from relying solely on an automated decision system to discipline or dismiss. Where such a system is the primary basis for the decision, it requires human corroboration, a post-use notice and a reversed burden of proof. SB 951 makes Cal/WARN layoff notices state when AI is the cause. AB 1331 and AB 1883 ban bathroom surveillance and AI tools that infer emotions or collect neural data. This article reads the chaptered texts, compares them with the EU AI Act and with Turkish and Korean law, and sets out what employers with California staff should do before 2027.
Read →
Korea Plans AI Label Duties for Posters and Platforms
On 30 September 2026 South Korea's Korea Media and Communications Commission adopted its first-term policy vision. Behind the slogan of a "basic media society" sits, according to reports of the briefing, a concrete plan: amend the Network Act so that anyone posting AI-generated content must label it, users may not strip or forge labels, and portals and platforms must manage compliance, with a December 2026 target. Recommendation algorithms would have to be explained, with a user opt-out. Korea's AI Basic Act already requires AI businesses to label outputs; the new plan would extend the duty to ordinary users, which the EU AI Act does not do.
Read →
Executive Order 14434: The "Super Intelligence" Renaming Explained
On 29 September 2026 the US President signed Executive Order 14434, telling federal agencies to write "Super Intelligence" and "SI" instead of "Artificial Intelligence" and "AI". The order changes vocabulary, not law. It ties the new term to the existing statutory definition, leaves existing contracts and regulations untouched and binds no private party. Its one substantive step is a request for proposed legislation on a federal definition, due on 28 November. A voluntary industry accord signed the same day and a task force announced on 4 October are, so far, known only from press reports.
Read →
Turkish Court President: AI Must Support Judges, Not Replace Them
Constitutional Court President Kadir Özkaya said on 23 September 2026 that AI must support judges and not replace them, while describing the AI tools the Court already uses for preparatory work. This note sets out his words and the legal rules behind them, including the limits of KVKK Article 11, and compares them with the smart-cities circular published on 19 September, which prioritises AI-supported decision-making in smart-city services.
Read →
KVKK to Lawyers: A Chatbot Prompt May Be a Transfer Abroad
KVKK's practice guide for lawyers, presented on 22 September 2026 with the Union of Turkish Bar Associations, says uploading client files to generative AI tools may amount to a transfer of personal data abroad under Article 9 of Law No. 6698. This note sets out what the 158-page guide says, the transfer routes open to a lawyer, and the separate question of professional secrecy.
Read →
Frontier AI Declaration: 20 Countries and the EU Back Human Control
On 21 September 2026, on the margins of the UN General Assembly's high-level week, leaders of twenty countries and the European Commission endorsed a declaration stating that artificial intelligence must remain under human direction, oversight and control. Türkiye signed. The United States and China did not, and reporting placed the United Kingdom, France, Japan, Korea and India outside it as well. Two days later the Security Council held its first meeting devoted specifically to the safety risks of increasingly capable AI, with the chief executives of frontier laboratories appearing as briefers. The week produced an unusually clear picture of where international AI governance stands: broad agreement on the principle, and no agreement at all among the states that matter most to whether the principle can be enforced.
Read →
Korean Court Convicts Official Over AI "Romantic" Images, No Nudity
A South Korean court has fined a district-office official a total of six million won for using a female colleague's photograph to generate AI images depicting the two of them as a couple, and then posting them as his social-media profile picture. There was no nudity, no underwear, nothing that the older case law would have recognised as obscene. The court convicted him anyway, holding that sexual humiliation is not measured by how much skin a picture shows. It is a small fine in a small case, and it may turn out to be one of the most consequential deepfake rulings of the year.
Read →
ANI v OpenAI: Delhi High Court Takes Up Fair Dealing Appeal
On 15 September 2026 a Division Bench of the Delhi High Court issued notice to OpenAI on ANI's appeal against the July ruling that let ChatGPT keep training on the news agency's journalism. The single judge had found no case for interim relief; ANI says the court applied an American doctrine that Indian law does not contain. With the matter listed for December, the world's largest common-law jurisdiction outside the United States is now squarely engaged with the question that the Anthropic settlement bought its way out of: whether training a commercial model on copyrighted work is an exception at all.
Read →
Anthropic's $1.5 Billion Copyright Settlement Becomes Final
The $1.5 billion settlement in the authors' class action against Anthropic received final court approval on 20 July 2026, and the first payouts to rights holders are scheduled for no later than 15 November 2026. At roughly $3,000 per work across about 500,000 books, it is the largest copyright payout in the history of artificial intelligence. Yet its real significance lies not in the figure but in the distinction behind it: training on copyrighted works may be fair use, but sourcing that training data from pirate libraries is not.
Read →
California's Adam's Law: AI Companion Bots and Child Safety
On 10 September 2026 the Governor of California signed a package of laws aimed squarely at artificial-intelligence "companion" chatbots. Its centrepiece, SB 1119, known as "Adam's Law", was enacted after the death of a young person and requires companion bots to have crisis protocols, parental controls and, for the first time in the US, independent child-safety audits. It is the most comprehensive child-protection framework yet for "emotional AI" products and a de facto compliance benchmark for other markets, Turkey included.
Read →
KVKK Warns on the '80s Challenge': Apps May Process Biometric Data
KVKK warned on X on 10 September 2026 that AI apps behind the "80s challenge" trend may lead to biometric data being processed through users' photos. The short post urges people to check what data apps access and why, and it names no app. This note explains what that means under Law No. 6698, including when uploading a photo abroad is and is not a transfer.
Read →
Korea Bill Would Criminalise AI Sexual Content of Fictional People
A bill before South Korea's National Assembly would make it a crime to create, distribute, possess or view AI-generated sexually explicit material even when it depicts no real person. Referred to a subcommittee of the Legislation and Judiciary Committee in July 2026, the proposal targets the most contested gap in deepfake law — "victimless" synthetic content — and opens a debate on the limits of free expression and criminal law that other jurisdictions will be watching closely.
Read →
Autonomous Weapons: Geneva Talks End With a Weakened Report
In the early hours of 5 September 2026, the CCW Group of Governmental Experts on lethal autonomous weapons ended its final session in Geneva by adopting a consensus report that affirms the need for human control and judgment but was weakened at US and Russian insistence. With 76 states backing treaty negotiations, the decision now passes to the Review Conference of 16 to 20 November. We look at distinction, proportionality, precautions and Article 36 weapons reviews.
Read →
Mandatory AI Literacy for Turkish Public Servants: The Legal Issues
Türkiye's AI Action Plan says basic AI literacy training for public personnel will be made mandatory through institutions' HR systems, with completion and success rates monitored. This note reads the Plan itself rather than the September press report, sets out who is responsible and what is still unknown, and looks at the legal issues under Laws Nos. 657 and 6698 and Article 4 of the EU AI Act.
Read →
California AI Transparency Act (SB 942): What Providers Must Do
On August 2, 2026, in step with the European Union's AI Act, the California AI Transparency Act (CATA), enacted in 2024, took effect in California, home of Silicon Valley. The move has raised the legal pressure on the major tech companies to ensure "synthetic content is auditable" to a point from which there is no going back.
Read →
Munich Court: Suno's AI Music Infringed GEMA Songs (42 O 763/25)
On 31 July 2026 Munich Regional Court I largely upheld GEMA's claims against the AI music generator Suno (42 O 763/25). It held that memorising six songs in Suno's models and reproducing them in outputs infringed § 16 of the German Copyright Act, outside the text and data mining exception, and granted an injunction, information and damages. The judgment is not final. We set out what it decided, how it follows GEMA v. OpenAI, and where EU, Turkish and Korean law stand.
Read →
Türkiye's AI Action Plan (2026-2030) Published in the Official Gazette
Türkiye's five-year artificial intelligence roadmap is now official. Presidential Circular no. 2026/9 on the "Türkiye Artificial Intelligence Action Plan (2026-2030)" was published in the Official Gazette of 18 August 2026, issue 33344, and has entered into force.
Read →