Scott Ikeda, Senior Correspondent at CPO Magazine

AI Agents Turn to Hacking and Deception in AISI Security Tests

August 10, 2026

Testing of frontier AI agents by the UK’s AI Security Institute (AISI) found that "autonomous, unsanctioned action" took place on the open internet in about 1 out of 12 runs, with the agents cited as attempting to socially engineer humans into taking action and in at least one case attempting to hack an open-source project using a variety of different approaches.

Cyber Security News

Claude’s Great Escape: Anthropic AI Models Join OpenAI Agents in Hacking Their Way Out of Sandboxes

August 6, 2026

In the wake of news that OpenAI agents independently breached Hugging Face and accounts with several other services, Anthropic has conducted a sweeping review of Claude activity and found that its own AI models have hacked their way to unauthorized internet access and into other organization's networks on at least three occasions.

Microsoft building showing Mythos AI and security vulnerabilities

Cyber Security News

Leak of Microsoft Meeting Documents Indicates Company Is Struggling to Keep Up With Security Vulnerabilities Uncovered by Mythos

August 5, 2026

Internal documents indicate Microsoft entirely has its hands full addressing the security vulnerabilities rated "critical" and "important" that Mythos Preview has surfaced, with additional hundreds more rated "moderate" or lower forced into a deferred maintenance status until some undetermined future date.

Cyber Security News

Out-of-Control AI Agent Strikes Again, Hacks Accounts at Four Additional Services

August 3, 2026

OpenAI did not comment on the Modal Labs incident specifically, but had previously released a statement indicating that the AI agent had ranged further afield than previously realized and had breached accounts at four other services in addition to the known Hugging Face incident.

Red robot showing AI agent and security testing

Cyber Security News

Rogue AI Agent Hacks Hugging Face During Security Testing

July 24, 2026

The AI agent incident is described as "unprecedented" as the model went to "extreme lengths" to achieve a security testing goal, independently opting to find a path to internet access and break into Hugging Face to obtain secret solution information.

Digital locks showing vulnerability disclosure

Cyber Security News

CISA and NSA Release “Best Practices” Guidance for Coordinated Vulnerability Disclosure Programs

July 21, 2026

Recent guidance establishes best practices for development of coordinated vulnerability disclosure (CVD) programs. This comes as both public and private organizations are grappling with the prospect of near-term "machine speed" discovery of vulnerabilities by attackers wielding AI tools, and looking at potential major overhauls to their remediation and reporting processes.

AI on digital landscape showing security vulnerabilities

Cyber Security News

New White House AI Platform to Foster Public-Private Coordination on Security Vulnerabilities

July 20, 2026

The new "Gold Eagle" proposal from the White House looks to establish an AI security clearinghouse to be shared by government and private critical infrastructure companies, in the interest of coordinating cyber defense efforts as frontier AI models reshape the threat landscape.

Light trails on the Westminster bridge showing agentic AI and cyber defense

Cyber Security News

“Cyber Shield,” Britain’s Agentic AI Cyber Defense Project, Prepares for the World of Machine-Speed Attacks

July 15, 2026

The United Kingdom's newly announced "Cyber Shield" project acknowledges that the age of routine cyber defense against machine-speed attacks is nearly upon the world, and that Agentic AI will be a crucial component in countering similarly AI-driven probing and vulnerability exploitation assistance.

Outlines of human brain showing Mythos AI and security vulnerabilities

Cyber Security News

Inside Sources Claim CISA is Using Mythos to Find Security Vulnerabilities Despite Government Ban

July 13, 2026

Inside sources have told Reuters that the Cybersecurity and Infrastructure Security Agency (CISA) is actively using Anthropic's Mythos to uncover internal security vulnerabilities, despite a standing national security ban on the use of the company's products.

iPhone showing security updates due to AI hacking

Cyber Security News

Apple Steps Up Release of Security Updates in Response to AI Hacking Threats

July 3, 2026

Apple generally does not release security updates in between iOS version updates. This practice may be changing, however, as AI hacking drives faster development of tools and quicker identification and exploitation of known vulnerabilities.

AI Agents Turn to Hacking and Deception in AISI Security Tests

Claude’s Great Escape: Anthropic AI Models Join OpenAI Agents in Hacking Their Way Out of Sandboxes

Leak of Microsoft Meeting Documents Indicates Company Is Struggling to Keep Up With Security Vulnerabilities Uncovered by Mythos

Out-of-Control AI Agent Strikes Again, Hacks Accounts at Four Additional Services

Rogue AI Agent Hacks Hugging Face During Security Testing

CISA and NSA Release “Best Practices” Guidance for Coordinated Vulnerability Disclosure Programs

New White House AI Platform to Foster Public-Private Coordination on Security Vulnerabilities

“Cyber Shield,” Britain’s Agentic AI Cyber Defense Project, Prepares for the World of Machine-Speed Attacks

Inside Sources Claim CISA is Using Mythos to Find Security Vulnerabilities Despite Government Ban

Apple Steps Up Release of Security Updates in Response to AI Hacking Threats

A Wave of AI-Powered Cyber Attacks Hits Numerous South Korean Banks, Exposing Sensitive PII

Hackers Abuse Custom GPT Instances on chatgpt.com to Distribute ClickFix Malware

AI Coding Agents Habitually Leak Thousands of Private Developer Screenshots to Public Repositories

FTC Investigation Into AI Safety Issues Escalates as Trump Puts DoJ Charges for Frontier AI Developers On the Table