EU Places ChatGPT Under Its Strictest DSA Oversight by December 2026

01EU Designates ChatGPT a ‘Very Large Online Search Engine,’ Requiring Compliance With the Strictest DSA Rules by the End of 2026

The European Commission has designated ChatGPT a “Very Large Online Search Engine” under the Digital Services Act, moving OpenAI’s chatbot into the European Union’s highest tier of oversight for major online services. The designation applies once an online platform or search engine reaches at least 45 million average monthly users in the EU—a threshold ChatGPT has now crossed.

That new legal status means OpenAI will be held accountable for mitigating risks connected to ChatGPT’s impact on minors and users’ mental health, as well as the spread of illegal content. The DSA is the EU’s framework for regulating major online platforms and search engines, with additional scrutiny applied to services whose scale gives them a large impact on citizens and society.

The Commission announced the ChatGPT designation alongside similar decisions for Reddit and Roblox, which were classified as “Very Large Online Platforms.” All three services must comply with the applicable rules by the end of December 2026.

The higher obligations include restrictions on targeting advertisements to minors. They also prohibit advertising based on sensitive personal characteristics such as a person’s sexual orientation, religion, ethnicity or political beliefs. Covered services must provide greater transparency about how their recommendation algorithms work.

For ChatGPT, the most consequential requirements concern the risks created by an AI service used at this scale. OpenAI will have to address how the chatbot may affect younger users and mental health, while also accounting for its potential role in distributing illegal material. The source does not describe the product, moderation or safety changes OpenAI will make, nor does it provide specific assessments of those risks.

Henna Virkkunen, the EU executive vice-president responsible for tech sovereignty, security and democracy, said the designations would hold ChatGPT, Reddit and Roblox to a higher standard of scrutiny and accountability because of their reach and societal impact.

The immediate change is regulatory: ChatGPT now has a formal status that carries enhanced obligations under the DSA. The practical changes experienced by users will depend on how OpenAI implements those requirements, which has not yet been disclosed. The firm deadline is the end of December 2026, when ChatGPT and the two newly designated platforms must be in compliance.

02US tightens access for foreign drones and robots, but September tariffs are unlikely to erase China’s manufacturing scale advantage

Washington tightened restrictions on foreign-made advanced robotic systems in July and August and imposed steep tariffs on imported drones and their components, citing national-security concerns. The drone tariffs take effect in September, while additional component tariffs are scheduled for 2027; the source does not specify the tariff rates.

The measures extend a broader effort to restrict foreign technology in strategically important industries. The Federal Communications Commission’s Covered List, created in 2021 to target communications and surveillance equipment from companies including Huawei, ZTE and Hikvision, later expanded to foreign-made drones and most recently to advanced robotic devices.

The restrictions may shield parts of the American market, but analysts and industry executives interviewed by TechCrunch said they do not directly overcome Chinese manufacturers’ advantages in production scale, supply-chain depth and cost. Chinese companies already compete in drones and humanoid robots at prices that many US and European rivals struggle to match.

Counterpoint Research estimated that 22,000 humanoid robots shipped worldwide during the first half of 2026, with the vast majority coming from Chinese manufacturers. The five largest makers by shipments—AgiBot, Unitree, Galbot, UBTECH and Leju Robotics—were all Chinese and together represented 86% of global shipments during that period. That figure describes the combined share of those five companies, not every Chinese manufacturer.

Scale can reinforce the cost advantage: lower prices put more robots into use, creating real-world data that can help improve the technology, while higher volumes can reduce costs further. Chinese manufacturers are also bringing more of the technology stack in-house and drawing on the country’s established manufacturing base.

Analysts expect tighter US access to push Chinese robotics companies toward Europe, Southeast Asia, Latin America and the Middle East, particularly in price-sensitive markets facing labor shortages. The outcome remains uncertain, but industry participants anticipate a more regional market: Chinese companies expanding through low-cost, high-volume production, while US and allied manufacturers concentrate on defense and critical infrastructure, where security requirements carry more weight. The source does not establish how much redirected capacity other markets can absorb or whether American manufacturers can scale production.

0398% of Claims Adjusters’ Workplace Reviews Mentioning AI Are Negative as Misclassification and Hallucinations Turn Automation Into Rework

Insurance claims adjusters open claims, collect details about accidents or losses, review supporting records and help determine payouts. Insurers increasingly use AI for those steps, from initial loss reports and damage estimates to medical-record summaries and automated payments. But adjusters who discuss AI in Glassdoor workplace reviews criticize it 98% of the time—the most negative response among the occupations examined by the platform.

The finding comes as insurers accelerate automation. Lemonade, an insurer built around bots and machine learning, said its AI Jim chatbot handled 96% of initial reports by the end of last year, while automated systems processed roughly 55% of all claims. Traditional insurer State Farm said it seeks to combine human and digital expertise, while giving employees more time to help customers.

Frontline workers say the technology can instead move work downstream. Ahmad Jackson, a former adjuster at a major insurer, said an AI system introduced for initial loss reporting sent a surge of misclassified claims to the wrong departments. Human adjusters then had to inspect and reroute them.

Jackson also encountered hallucinations—plausible-sounding but false AI output—in claim summaries. When he unknowingly repeated those errors to claimants or their lawyers, he said, their anger fell on him rather than the system. He later changed employers, although he still considers AI useful for limited administrative tasks such as extending a customer’s rental-car booking.

Other errors can affect payments. Insurance consultant and former adjuster Sandy Avina said a smudge on a legal document can prompt an AI hallucination, while an omitted detail in a medical summary can also produce incorrect information or a wrong payout. Customers may not know automation caused the mistake and instead blame the adjuster handling their case.

The backlash coincides with a shrinking job market. The US Bureau of Labor Statistics previously projected 18,900 fewer adjuster jobs over a decade, a 5% decline, and identified technology as one important factor. BLS data also showed sector employment falling 21% from May 2025 to May 2026, while Glassdoor said entry-level postings had dropped 50% since 2025.

Those figures show that adjusters face rapid employment pressure, but they do not establish AI as the independent cause. The available account also does not disclose Glassdoor’s full methodology or sample size, or insurers’ actual error rates.