AI firms, science & scrutiny
OpenAI released an early model aimed at speeding drug discovery, putting it in direct competition with rivals such as Google. (latimes.com) At the same time some OpenAI shareholders are reportedly questioning Sam Altman's ability to lead the firm to an IPO and there are reports his personal investments have blurred lines between company and personal interests. (the-decoder.com)(gizmodo.com)(livemint.com)(news.futunn.com) Meanwhile, Anthropic’s new cybersecurity-focused model has attracted attention from the U.S. administration as firms try to align commercial AI work with government priorities. (theverge.com)
Drug discovery is a long filtering process: researchers test huge numbers of molecules to find a few worth taking into the lab. OpenAI said on April 16 it is now offering an early model, GPT-Rosalind, built to help with that first pass across biology, chemistry and genomics. (openai.com) OpenAI said GPT-Rosalind is a research preview available in ChatGPT, Codex and the application programming interface for qualified customers, and that it can connect to more than 50 scientific tools and data sources through a Codex plugin. The company said customers using it include Amgen, Moderna, Thermo Fisher Scientific and the Allen Institute. (openai.com) The company is entering a field other large AI firms already treat as a proving ground for commercial science tools. The Los Angeles Times reported April 17 that OpenAI’s launch puts it into more direct competition with Google in AI systems aimed at speeding early-stage drug research. (latimes.com) OpenAI framed the pitch around time: it said a new drug in the United States typically takes 10 to 15 years to move from target discovery to regulatory approval. Its argument is that faster literature review, hypothesis generation and experiment planning could improve decisions before companies spend money on wet-lab work and clinical trials. (openai.com) That product push is landing as OpenAI is also under investor scrutiny over leadership and governance ahead of a possible stock-market debut. OpenAI said on March 31 that it closed a $122 billion funding round at a post-money valuation of $852 billion. (openai.com) The Wall Street Journal reported, and follow-up reports summarized, that some shareholders are questioning whether Sam Altman should lead OpenAI through a planned initial public offering at roughly that valuation. The same reporting said Altman’s personal investments are hard for outsiders to map against OpenAI’s business interests, raising conflict questions the company would need to answer more fully as a public issuer. (wsj.com) (the-decoder.com) A separate report, cited by Reuters pickups last week, said Chief Financial Officer Sarah Friar had raised concerns internally about whether OpenAI was ready for an initial public offering as early as the fourth quarter of 2026. Those reports also said she questioned the risks around Altman’s spending plans and the company’s cash needs. (economictimes.indiatimes.com) (thehindu.com) Anthropic is moving on a parallel track, but in cybersecurity instead of life sciences. TechCrunch reported April 14 that co-founder Jack Clark said Anthropic had briefed the Trump administration on Mythos, a model the company has not released publicly because of what it described as powerful cyber capabilities. (techcrunch.com) Clark said Anthropic still wanted the government to know about systems that touch national security even while the company was fighting a contracting dispute with the Department of Defense. That puts two of the biggest AI firms in the same place at once: selling specialized models to industry while trying to show Washington that their most sensitive systems fit government priorities. (techcrunch.com)
