AI firms under pressure
Leading AI companies are funding policy papers, think-tanks and research institutes while spending millions on lobbying even as they publicly call for oversight. Regulators and governments are reacting: the EU is weighing whether to treat ChatGPT as a "large online platform" under the Digital Services Act after disclosures of more than 45 million users, and U.S. officials convened CEOs over safety concerns tied to Anthropic’s model 'Mythos.' Separately, OpenAI disclosed a security issue involving a third-party developer tool called Axios but said user data was not accessed. ( )
Artificial intelligence companies that spent years asking governments to regulate them are now facing scrutiny over how they try to shape those rules themselves. European and U.S. officials moved this week as new reporting detailed lobbying, policy funding and security concerns around the industry. (enterpriseai.economictimes.indiatimes.com, timesofindia.indiatimes.com, openai.com) The European Commission said on April 11 it was assessing whether ChatGPT should fall under stricter Digital Services Act rules after OpenAI disclosed more than 45 million monthly users in the European Union. That threshold can trigger extra obligations for very large services, including outside audits and risk controls. (enterpriseai.economictimes.indiatimes.com, euractiv.com) In Washington, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell joined calls with major bank chiefs and technology executives after concerns about Anthropic’s Claude Mythos Preview. CNBC reported Anthropic limited the rollout because of fears the model could be misused to find and exploit software flaws. (timesofindia.indiatimes.com, cnbc.com, nextgov.com) Those moves landed as AI companies widened their influence campaigns outside formal regulation. The Guardian reported on April 12 that firms were backing policy papers, think tanks and research institutes while also increasing lobbying and public messaging around artificial intelligence. (theguardian.com, goodmorningamerica.com) OpenAI published a 13-page policy paper on April 6 that proposed ideas including a public wealth fund, taxes tied to automated labor and shorter workweeks. Anthropic separately launched the Anthropic Institute, presenting it as a venue to study artificial intelligence’s economic and social effects. (techcrunch.com, theguardian.com) The money around AI policy is spreading beyond white papers. Politico reported on April 1 that Common Sense Media was asking top AI companies for as much as $100 million each to help fund a new safety effort, while ABC News reported that AI-linked groups were already pouring millions into the 2026 midterms. (politico.com, goodmorningamerica.com) Companies say they are trying to build safeguards as the technology spreads quickly. Anthropic said it briefed senior U.S. officials on Mythos Preview before any external release, and OpenAI said its Axios-related incident was part of a broader industry compromise rather than a breach of its own core systems. (nextgov.com, openai.com) OpenAI said on April 10 that it found a security issue involving a third-party developer tool called Axios and was protecting the process used to certify legitimate macOS apps. The company said it found no evidence that user data, systems or intellectual property were accessed, and told macOS users to update their apps. (openai.com, tech.yahoo.com) The next test is whether governments treat these companies more like ordinary software vendors or like infrastructure that can move markets, shape public debate and create new security risks. Brussels is still assessing ChatGPT’s status, and Washington has already shown it will pull in chief executives when a model appears to cross a new line. (enterpriseai.economictimes.indiatimes.com, cnbc.com)
