Anthropic, an artificial intelligence (AI) firm, has warned that the world could lose control of advanced AI systems after internal data showed its Claude models had surpassed human engineers on some coding tasks. The company published a report on June 4 showing Claude wrote over 80% of the code merged into its systems in May 2026. The findings prompted Anthropic to call for a global plan to slow AI development before it begins improving itself without human input.
Introducing TipRanks MCP for Agents
High conviction NVDA bears now have this Tradr ETFClaude Hits Superhuman Speeds on Core Tasks
In a blog post titled “When AI Builds Itself,” Anthropic researchers Marina Favaro and Jack Clark shared internal data showing how quickly Claude Code models are advancing. On the hardest open-ended tasks, Claude’s success rate rose from 26% to 76% in six months.
The company also said engineering teams are now shipping about 8x as much code per day as in 2024. It also cited a case where Claude resolved over 800 bugs, reducing a class of API errors, a work that an engineer estimates would have taken a human four years to complete.
The firm added that Claude’s performance in experimental workflows has improved sharply. It moved from about a 3x speedup in May 2025 to around 52x by April 2026. Anthropic said Claude now outperforms skilled human researchers on clearly defined tasks.
Anthropic Warns of AI Control Risks, Calls for Global Action
Anthropic stated that its findings have implications that warrant closer attention. The company said Claude is accelerating AI development faster than expected.
If this continues and grows, Anthropic warns it could lead to systems that can design and build more advanced successors on their own without human help. The firm described this process as a “recursive self-improvement.”
Anthropic acknowledged that these advances could bring major benefits to areas like medicine, tech, and the economy. However, the company warned that safety risks could rise and the chance of losing control over these AI systems would also increase.
The firm said its research institute will study the risks of powerful AI systems, including those that may one day improve themselves. It added that the world should have the option to slow or briefly pause AI development to allow safety research and policy to keep pace.
The firm also suggested that countries may need a shared agreement and a system to verify compliance. Anthropic said self-improving AI is not assured and has not happened yet, but it could arrive sooner than many governments expect.
Is Anthropic on the Stock Market?
Anthropic is still a private company and does not have shares listed on any public stock exchange. Although the firm filed confidential initial public offering (IPO) paperwork in late May 2026, it has not set an official listing date. Investors seeking exposure to the AI industry can consider publicly traded companies such as Nvidia (NVDA), Micron (MU), and Microsoft (MSFT). Visit the Tipranks Stocks Comparison Center to get analysts’ ratings and forecasts for these stocks.
