Anthropic, the AI firm known for the Claude chatbot and its safety-oriented approach, seems to be adjusting its safety protocols to stay competitive. The company revealed its revised responsible-scaling policy, a set of internal rules aimed at averting potential AI risks like large-scale cyberattacks.
While the updated guidelines emphasize the need for a “strong argument that catastrophic risk is contained” during AI development, they now allow progress to continue “until and unless we no longer believe we have a significant lead,” indicating that development may proceed if the company doesn’t perceive itself as ahead of competitors.
Anthropic cited a shift in focus from AI safety to economic potential in the U.S. as the reason for this policy change. The company mentioned that despite advancements in AI capabilities, governmental actions on AI safety have been sluggish, with more emphasis placed on AI competitiveness and economic growth rather than safety discussions at the federal level.
The alteration in Anthropic’s safety framework coincides with a situation where the Pentagon threatens to terminate contracts unless Anthropic’s technology can be used for all legal military purposes. However, Anthropic asserts that this safety guideline adjustment is unrelated to the Pentagon’s ultimatum.
Founded in 2021 by former OpenAI employees concerned about safety priorities, Anthropic has always emphasized safety as its top priority. CEO Dario Amodei has expressed apprehensions about the negative impacts of AI, affirming safety as the company’s primary focus in interviews.
The company’s recent safety policy update includes commitments to enhancing transparency and accountability by regularly publishing safety reports and goals. Despite Anthropic’s safety-first reputation, Heidy Khlaaf, from the AI Now Institute, criticized the company for not adequately addressing potential harms from current AI technologies, such as chatbot errors.
Although Anthropic has been associated with safety in the past, Khlaaf believes the company is now shedding its safety-centric image to align with market demands. Amid escalating competition among top AI firms like Anthropic, OpenAI, and Google, safety considerations are challenged by U.S. governmental emphasis on AI development and competitiveness.
The lack of clear regulations in both the U.S. and Canada poses challenges for AI companies like Anthropic to prioritize safety over economic gains. The absence of broad AI regulation since the failure of Canada’s Artificial Intelligence and Data Act in 2025 reflects a reluctance to impose stringent rules on AI development.
Anthropic’s safety policy revision coincides with pressure from the Pentagon, following a lucrative deal allowing the use of its AI technology for military purposes within specified guidelines. The Pentagon’s directive to allow broader use of Anthropic’s technology in military applications has raised concerns about potential misuse and ethical implications.
Anthropic has reaffirmed its stance against the use of its technology in autonomous weapons and mass surveillance systems, despite the Pentagon’s demands. The dispute between Anthropic and the government primarily revolves around usage policies rather than scaling policies according to company statements.
As the deadline looms, Anthropic remains firm in its position, emphasizing its commitment to ethical use of technology and willingness to switch providers if necessary. The company’s principled stance against certain military applications underscores its dedication to responsible AI development.
