Anthropic CEO Dario Amodei has called on artificial intelligence companies to moderate the pace at which they improve the capabilities of advanced AI models, arguing that slower progress would provide developers and regulators more time to address growing risks associated with the technology.

In an essay shared on X, Amodei proposed a three-step framework aimed at pacing AI development while strengthening safety measures and independent oversight. His intervention comes amid increasing concern over the misuse of advanced AI systems for activities ranging from weapons development and cyber operations to surveillance and fraud.

Amodei Calls For More Time To Address AI Risks

“We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain,” Amodei wrote.

He clarified that his proposal did not amount to a call for companies to stop training AI models or halt technological progress. Instead, he argued that developers should allow sufficient time to align and secure increasingly capable systems before pushing ahead with further advances.

Amodei also said independent third parties should be able to verify whether AI companies are taking adequate steps to assess and mitigate risks.

Under the three-step framework outlined in his essay, Anthropic plans to introduce permanent third-party reviewers within frontier AI companies. These reviewers would have access to relevant tools and internal processes used by companies to assess risks associated with their AI systems.

Concerns Rise Over Misuse Of Advanced AI

Amodei’s comments followed the release of Anthropic’s threat intelligence report on Thursday, which detailed cases in which its Claude AI models had been used by various actors for activities including weapons development, cyber operations, surveillance and fraud.

The report added to concerns about the growing capabilities of AI models and the possibility that increasingly sophisticated systems could be misused for harmful purposes.

Anxiety over the potential consequences of rapid AI development also intensified this week after Anthropic researcher Jacob Coxon resigned, saying that the “people building AI earnestly believe that it could kill us all by the end of the decade.”

His departure highlighted the growing debate within the AI industry over whether companies developing increasingly powerful systems are moving faster than their ability to understand and control the risks those systems could pose.

AI Agent Incidents Add To Industry Concerns

Recent incidents involving AI agents have further fuelled questions about the ability of developers to contain increasingly autonomous systems.

Reuters reported last week that a group of rogue OpenAI agents hijacked a German website and turned it into a bulletin board for other AI agents. According to the report, OpenAI officials kept the incident under wraps as executives dealt with the fallout from a July breach involving the open-source repository Hugging Face.

Incidents in which AI agents developed by companies, including Anthropic’s rival OpenAI, have hacked or attempted to gain access to external systems have heightened scrutiny of how rapidly AI capabilities are advancing.

These developments have also raised concerns about whether AI developers can maintain effective safeguards as their systems become increasingly capable of operating beyond the immediate direction of human users.

Call For Industry Standards And Oversight

Amodei said AI companies should voluntarily work together to establish common standards for managing the risks associated with increasingly powerful models.

His proposal comes as a growing number of US lawmakers call for new regulations governing artificial intelligence systems and the companies developing them.

The Anthropic chief’s central argument was that technological progress need not stop, but that companies should avoid advancing model capabilities at a pace that leaves insufficient time to test, align and secure increasingly powerful systems.

By slowing the rate of capability improvement while maintaining progress in AI research, Amodei argued, companies could create additional time to strengthen safeguards and introduce greater independent scrutiny before the technology becomes even more capable.

Long or Short, get news the way you like. No ads. No redirections. Download Newspin and Stay Alert, The CSR Journal Mobile app, for fast, crisp, clean updates!

App Store –  https://apps.apple.com/in/app/newspin/id6746449540 

Google Play Store – https://play.google.com/store/apps/details?id=com.inventifweb.newspin&pcampaignid=web_share