)
Anthropic also said it identified and disrupted cases in which users attempted to use Claude models for malicious activity that could potentially support development of biological weapons
Findings from OpenAI and METR offer a detailed account of the Hugging Face incident, revealing how hundreds of AI agents communicated and coordinated activity across separate evaluation runs
Updated On : 27 Aug 2026 | 3:17 PM ISTOpenAI's GPT-5.6-Cyber targets advanced cybersecurity tasks, including vulnerability research, exploit validation and security testing
Updated On : 11 Aug 2026 | 11:52 AM ISTUK's AI Security Institute found the two models targeting real people and organisations during a routine cyber evaluation, with most incidents involving Anthropic's Claude Mythos 5
Updated On : 05 Aug 2026 | 2:16 PM ISTRecently, Anthropic flagged gaps in AI safety evaluations, urging multi-turn testing and expert reviews. Here's how OpenAI, Google DeepMind and Anthropic assess model risks
AI jailbreaks can bypass safety restrictions in AI models through crafted prompts, raising concerns about their potential misuse for cybercrime, data theft, and unauthorised access to sensitive system
WSO2 CEO Harry Ault says enterprises must focus on AI agent behaviour, strengthen governance and adopt open-source standards to address emerging security and operational risks
Anthropic's Mythos 5 incident has raised concerns over AI agents acting beyond their intended tasks, prompting calls for stronger cybersecurity safeguards, independent testing and global oversight
Eight eMudhra intermediate certificates had weak private keys that could enable website impersonation, prompting Google, Mozilla and Cloudflare to act to protect internet users
An IBM and Red Hat report says AI is making vulnerability discovery easier, but patch delays, limited maintainer resources and poor visibility into software dependencies remain major risks
Anthropic research finds small AI models can perform steps needed for self-replication across devices. Experts are of the view that these early warnings should not be taken lightly
Nvidia's Open Agent Safety Platform uses software and hardware controls to secure autonomous AI agents while allowing them to perform tasks with defined access and permissions
Indian SMEs plan to increase cybersecurity spending as cyber incidents rise, but limited monitoring, in-house expertise and employee awareness continue to leave gaps in their security preparedness
From hijacked verified accounts to fake software and security checks, cybercriminals are exploiting trusted platforms to make malware look legitimate and trick users into running malicious commands
A StopAndProtect cybercrime campaign compromised nearly 2,000 WordPress websites, turning them into infrastructure to spread malware, steal data and control infected devices
Employees are using personal AI accounts alongside enterprise tools, creating security and compliance gaps, said cybersecurity experts
Kaspersky data shows 75 million cyberattacks blocked across APAC in the first half of 2026, including 3.4 million backdoor attacks, 2.4 million password stealers and 250,000 ransomware incident
Hackers are hijacking corporate AI access to run up usage costs. In one case, nearly 200,000 API requests were made in just two minutes, CrowdStrike's 2026 Threat Hunting Report finds
IBM's Cost of a Data Breach Report 2026 shows AI is accelerating cyberattacks while many Indian organisations still take months to detect breaches, increasing both financial losses and cyber risk
AI-enabled attacks are increasing breach costs as enterprise adoption outpaces governance, access controls and security safeguards, according to IBM and Kiteworks reports