Anthropic PBC, which has been focusing more on artificial intelligence safety than its peers for many years, has now relaxed its commitment to its "safety guardrails." This is one of the most dramatic policy shifts in the AI ​​industry to date, and it also reflects that a group of startups that once focused on helping humans are turning their attention to profit and success.

In the "Responsible Expansion Policy" released in 2023, the company stated that it will postpone the development of AI that may cause danger. But Anthropic has updated its rules in a blog post published on Tuesday, saying it will no longer do so if the company believes it does not have a significant advantage over its competitors.

Anthropic wrote in the article, "The policy environment has shifted to prioritizing AI competitiveness and economic growth, while at the federal level, security-oriented discussions have yet to gain substantial momentum."

The adjustment comes as a surprise because Anthropic has long tried to differentiate itself from its peers on its security stance as it competes with OpenAI, Alphabet's Google and Elon Musk's xAI for dominance in what many regard as a revolutionary new technology. Anthropic CEO Dario Amodei, who previously worked at OpenAI, left the company in part due to concerns that OpenAI focused more on commercialization and speed than security.

Both companies are now pushing to go public as soon as this year to capitalize on investor enthusiasm for AI. Anthropic was recently valued at $380 billion, while OpenAI raised funding at a valuation of more than $850 billion.

A spokesperson for Anthropic said, "From the beginning, we have said that the speed of AI development and the uncertainty in the field require us to quickly iterate and improve policies."

Earlier this month, senior security researcher Mrinank Sharma said he was leaving Anthropic. "I continue to find myself taking stock of the situation we find ourselves in," he wrote in a letter to colleagues posted to the social platform X.

“The world is in danger,” he wrote, “and not just from AI or biological weapons, but from a series of interconnected crises happening right now.”