Full Summary
This Sunday morning, a US national security order has forced Anthropic to suspend access to two of its advanced AI models, Fable 5 and Mythos 5, for all foreign nationals, even those within the company. Both The Korea Times and Cryptonews.net confirm this directive came just three days after Fable 5's public launch. The government believes a method exists to "jailbreak" Fable 5, potentially aiding hacking, though Anthropic disputes this. Meanwhile, the FBI, in collaboration with Google and Black Lotus Labs, has dismantled "Outsider Enterprise," a massive Chinese phishing operation. BleepingComputer reports this AI-powered ring used thousands of fake websites to steal over 3.8 million credit card records, causing an estimated $1.9 billion in losses since 2023. As part of "Operation Riptide," the FBI seized servers and approximately $100,000. In other AI security news, Zcash founder Zooko Wilcox reports Anthropic's Claude Mythos found no serious vulnerabilities in the Zcash protocol, a positive note following a recent forgery bug discovery assisted by Claude Opus 4.8. However, a new report in Tech Times suggests AI prompt injection might be a permanent flaw, not a fixable bug, inherent in AI agent design. This means hostile instructions can be smuggled into an AI, gaining the same authority as legitimate commands. This surge in AI-related security concerns impacts everyone. The potential for advanced AI tools to be weaponized, coupled with fundamental design flaws, means your digital interactions and personal data face increasingly sophisticated threats.