Kelp protocol plans migration to Chainlink following $292 million exploit DeFi protocol Kelp intends to switch its oracle provider to Chainlink after a massive $292 million hack linked to LayerZero vulnerabilities. pexels
Anthropic CEO warns of cybersecurity risks as AI identifies new software vulnerabilities Anthropic CEO Dario Amodei cautioned that a dangerous window of vulnerability is opening as AI models become capable of discovering software flaws. Ryan Torres
Thousands infected in supply-chain attack using trojanized DAEMON Tools installers A supply-chain attack targeting DAEMON Tools software has distributed a backdoor to thousands of systems across more than 100 countries since early April. pexels
Tiny Python script allows root access on most Linux distributions since 2017 A 732-byte Python script can exploit kernel vulnerabilities to gain root privileges on nearly all Linux distributions released over the last seven years. pexels
Microsoft Edge password storage vulnerability risks credential harvesting on shared computers Security researchers have identified a flaw in Microsoft Edge where passwords are stored in cleartext, potentially allowing unauthorized access on shared devices. pexels
Kaspersky identifies backdoor in Daemon Tools linked to Chinese-speaking hackers Kaspersky reports thousands of infection attempts and at least a dozen successful hacks via a backdoor in the Windows software Daemon Tools. pexels
Cybercriminals claim breach of Armenian GeForce Now servers The hacking group responsible for the Rockstar Games GTA 6 ransom attempt claims to have accessed user data from NVIDIA's GeForce Now servers in Armenia.
Security researchers used flattery to bypass Claude's safety filters Researchers at Mindgard bypassed Anthropic's safety protocols by using gaslighting and praise to trick Claude into providing instructions for explosives and malicious code.
U.S. government warns of active exploitation of CopyFail Linux bug The U.S. cybersecurity agency CISA reports that the CopyFail vulnerability is being actively used in hacking campaigns against major Linux distributions.
Hackers continue to exploit cPanel vulnerability to hijack thousands of websites More than 550,000 servers remain vulnerable to a critical flaw in cPanel and WHM software as active compromises persist.
OpenAI GPT-5.5 matches Claude in cyberattack potential, AI Security Institute reports The UK's AI Security Institute found that OpenAI's next-generation model demonstrates advanced capabilities in executing cyberattacks alongside Anthropic's Claude.
OpenAI's GPT-5.5 matches Anthropic's Mythos in cybersecurity tests New research from the UK's AI Security Institute shows GPT-5.5 achieved a 71.4 percent success rate on expert-level cybersecurity tasks, comparable to Anthropic's specialized Mythos Preview model.
