AI firms under pressure
Investors are questioning OpenAI’s $852 billion valuation as the company shifts toward enterprise customers, raising doubts about the sector’s durable business models. At the same time, Anthropic is publicly opposing an Illinois AI liability bill that OpenAI backed, and OpenAI has limited release of GPT‑5.4‑Cyber to trusted companies for vulnerability‑finding use cases. (reuters.com) (wired.com) (nytimes.com)
OpenAI’s latest week has turned into a three-front test of confidence: valuation, regulation, and product access. (channelnewsasia.com) On April 14, Reuters reported that some OpenAI backers are questioning the company’s $852 billion valuation as it shifts toward enterprise customers. The company raised $122 billion in March, but investors cited two product roadmap redraws in six months as competition intensified from Google and Anthropic. (channelnewsasia.com) The Financial Times, as quoted by Reuters, said some investors worry the strategy change could leave OpenAI exposed even as it prepares for a possible initial public offering in 2026. OpenAI pushed back, telling Reuters its March fundraising was oversubscribed and reflected “strong conviction” in its direction and long-term value. (channelnewsasia.com) At the same time, OpenAI and Anthropic have split publicly over an Illinois bill, Senate Bill 3444, that would limit when AI labs can be held liable for catastrophic harms caused by misuse of their systems. Anthropic told Wired it opposes the bill and has lobbied Illinois lawmakers to rewrite it or stop it. (europesays.com) Wired’s reporting says the bill would shield a lab from liability for large-scale harms, including mass casualties or more than $1 billion in property damage, if the lab published its own safety framework. Anthropic’s Cesar Fernandez said the proposal would give companies a “get-out-of-jail-free card,” while OpenAI argued the bill could reduce serious risks while keeping the technology available to Illinois businesses. (europesays.com) That fight lands as AI companies are spending more time in statehouses and more money proving they can sell durable products, not just fast-growing chatbots. Reuters said one early OpenAI backer questioned why a company with ChatGPT’s scale was talking so much about enterprise software and coding. (channelnewsasia.com) OpenAI also used April 14 to announce GPT-5.4-Cyber, a version of GPT-5.4 tuned for defensive cybersecurity work rather than general release. The company said it is expanding its Trusted Access for Cyber program to thousands of verified defenders and hundreds of teams protecting critical software. (openai.com) OpenAI said the model is “cyber-permissive,” meaning it is designed to help legitimate security teams find and fix vulnerabilities faster while access is gated with identity checks and other safeguards. The company said it has been building cyber-specific safeguards since 2025 and started evaluating model cyber capabilities in 2023. (openai.com) Taken together, the past two days show how the biggest AI labs are being judged on three clocks at once: quarterly revenue, legislative rules, and whether their most capable systems can be released beyond a narrow circle of trusted users. (channelnewsasia.com)
