Anthropic changes safety policy amid intense AI competition
When Anthropic launched years ago, the company wanted an industry-wide "race to the top" in artificial intelligence, instead of a race to the bottom in pursuit of customers and market dominance that would inadvertently lead to catastrophic safety risks.
So Anthropic adopted safety principles and policies that it hoped it competitors would also implement. In some instances, companies, including Google and OpenAI, did, according to Anthropic. Still, Anthropic's hopes didn't "pan out" as the company hoped, according to a blog post it published Tuesday.
The post announced that Anthropic, the maker of the AI chatbot Claude, is altering key safety practices to meet what it views as present-day challenges.
Specifically, Anthropic will no longer automatically pause model development if it could be considered dangerous; instead, it will consider its competitors' actions and whether they release models with similar capabilities. Previously, Anthropic committed to safeguards that would reduce its models' absolute risk, regardless of whether other AI developers did the same.
"The policy environment has shifted toward prioritizing AI competitiveness and economic growth, while safety-oriented discussions have yet to gain meaningful traction at the federal level," the company wrote. "We remain convinced that effective government engagement on AI safety is both necessary and achievable, and we aim to continue advancing a conversation grounded in evidence, national security interests, economic competitiveness, and public trust. But this is proving to be a long-term project—not something that is happening organically as AI becomes more capable or crosses certain thresholds."
Though Anthropic said it aims to continue leading on safety, its latest decision reflects the breakneck speed at which competitors are releasing new models.
Anthropic has also been under intense pressure this week by the U.S. Defense Department, which is pressing the company to allow the military to use its AI tools for any purpose, including mass surveillance or the deployment of autonomous weapons without human oversight.
Anthropic has yet to relent on those points in contract negotiations with the Defense Department, reportedly stirring the ire of Defense Secretary Pete Hegseth, who threatened to sever the company's relationship with the military, Axios reports.
Anthropic has participated in an AI pilot program for military-related imagery analysis, along with Google, OpenAI, and xAI, according to the New York Times. Though Claude has been the only chatbot working on the government's classified systems, a Pentagon official said Anthropic could be replaced by another firm.
Disclosure: Ziff Davis, Mashable’s parent company, in April 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.