Anthropic's newly released Claude Opus 5 introduces advanced cyber capabilities gated behind safety filters, but security researchers warn those guardrails may not hold against determined attackers. DeFi protocols are considered especially vulnerable: the sector has already suffered over $840 million in exploits in 2025 alone. Unlike traditional hackers, an AI-assisted attacker could autonomously scan smart contract code, identify vulnerabilities, and execute multi-step exploits at machine speed, compressing attack timelines from days to minutes. The concern is not hypothetical — previous AI models have already demonstrated the ability to identify real-world contract flaws in controlled tests. If safety filters are bypassed or jailbroken, the next major DeFi breach could be carried out at a scale and speed that outpaces any human defensive response.
— Sponsored —
Trade smarter on BYDFI
Get a bonus on your first deposit — from $50 at $100, up to $2,000 at $20k. 200x leverage, 600+ perpetuals, deep liquidity.