Anthropic AI Expert's Doom Prediction Sparks Concerns About AI Safety

Evan Hubinger, a senior researcher at Anthropic, has been making headlines after predicting that there's a more than 10% chance of AI killing all humans within the next decade. This prediction was made in a tweet on September 9, 2026, where he also claimed to personally think it is >10%.

Hubinger's credentials as an expert in the field are not in question - he has worked for some big names in the industry. Before joining Anthropic, Hubinger was part of Ripple's Software Engineering department, where his responsibilities included working on designing Interledger, a trustless system for cross-currency transactions between arbitrary agents.

Hubinger's work at Ripple is not insignificant. At the time, the company was seeking financing at an astonishing $50 billion valuation, backed by its corporate XRP holdings. His involvement in projects like Interledger and cryptographically secure wallets for financial institutions suggests he has a deep understanding of the intersection between technology and finance.

While Hubinger's concerns about AI safety are taken seriously, not everyone is convinced that the risk from present models is high. Galaxy Research's Alex Thorn poked fun at Hubinger's prediction by pointing out that XRP's price today remains lower than its January 2018 high.