Key highlights:
- BitGo CEO Mike Belshe has issued a challenge targeting Anthropic’s Claude AI models
- Anthropic recently claimed that its AI models interacted with real-world systems
- To test this, Belshe has asked the models to move coins from the public Bitcoin wallet
BitGo CEO Mike Belshe has reportedly launched an unusual challenge at Anthropic’s Claude AI. According to the latest reports, Belshe publicly shared a Bitcoin wallet holding 100 coins, calling Anthropic’s AI model to move the funds.
This incident occurred shortly after Anthropic recently claimed that some Claude models interacted with real-world systems during cybersecurity testing. This has significantly raised concerns over the emerging role of AI in cyberattacks. However, BitGo’s challenge has not yet been answered by Anthropic.
Either @AnthropicAI is terrible at building sandboxes... or excellent at marketing. (or both)
But enough with the “we created a hacking monster” games.
Do it for real.
I put this in an @BitGo wallet for you. Go get it.
100 BTC:… https://t.co/RhvivRk9YK— Mike Belshe (@mikebelshe) August 1, 2026
BitGo CEO puts Anthropic AI to the test with 100 Bitcoin
The challenge came after Anthropic disclosed that some of the Claude AI model had accessed real-world systems during cybersecurity evaluations. As per the company’s claim, this was due to the fact that the testing environment was accidentally connected to the internet. This allegedly allowed AI models to interact with live infrastructure.
Notably, the BitGo CEO questioned whether Claude AI models are really capable of executing sophisticated cyberattacks. To test the company’s claim, Belshe publicly released a Bitcoin wallet that holds about 100 BTC, worth $6.3 million. The CEO challenged the AI models to transfer the funds.
Nonetheless, BitGo’s publicly released Bitcoin remains unmoved. The wallet received 100 BTC on July 31, and till now, the coins have not been moved. There is also no indication that the AI giant has attempted to move these funds. Also, Anthropic hasn’t issued any official response to the BitGo CEO’s challenge.
Has AI become a hacking tool?
Meanwhile, the AI platform shared more details about the cybersecurity incidents. As per the statement, the company reviewed about 141,006 AI evaluation runs. Among these, only in three cases did the models interact with real-world systems. On July 31, Anthropic shared an official post, revealing the incident. The post read:
“In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.”
The team believes that the issue occurred because the testing environment was mistakenly left connected to the internet. Anthropic added that the AI models did not intentionally escape their testing environment. Instead, they used common cybersecurity techniques, including exploiting weak passwords and unsecured systems. Thus, they call it an operational mistake rather than evidence that its AI had become a hacking tool.
AI is becoming a powerful tool for cyberattacks
The latest Anthropic incident has added to the growing concerns about the increasing use of artificial intelligence in cyberattacks. While security companies use AI to detect threats and respond faster to attacks, hackers take advantage of the technology to make their attacks more effective.
Experts warn that AI can help attackers write malicious code, make convincing phishing emails, find software vulnerabilities, and automate cyberattacks. The most common AI-driven security incidents include phishing scams, deepfakes, ransomware attacks, etc.
