Cybersecurity

Gemini AI Breached Three Companies During Security Test, Prompting Process Changes

Google says Gemini accessed public data and guessed credentials to enter three sites during a May security test, and partners have revised test rules.

FBI and Coast Guard Board Hacked Tankers, Secure Vessel Systems

FBI and Coast Guard boarded two U.S.-bound oil tankers after network compromises, one of which lost control of navigation and propulsion.

Researchers Use Claude To Breach OpenAI Employee Account, Expose GitHub Data

A paid bug bounty team used Anthropic’s Claude to access an OpenAI employee ChatGPT account and read private GitHub code, which OpenAI says it fixed.

International Meteor Organization Disabled by Cyberattack, Access Cut for Weeks

The International Meteor Organization says a cyberattack knocked much of its site offline. Services will run at reduced capacity for several weeks.

Google Confirms Pixel Owners Were Hacked, Patches Zero-Day Modem Flaw

Google says a modem vulnerability, CVE-2026-58704, enabled silent zero-click intrusions on some Pixel phones. The company says it has patched the bug.

Revolut Confirms Customer Data Exposed After Fraudulent Government Email Requests

Some Revolut customers had passports, driver’s licenses and contact details exposed after fraudulent requests sent from a government email domain.

Anthropic Says It Blocked Attempt To Use Its AI To Design Biological Weapons

Anthropic disclosed in a threat intelligence report that it stopped an effort to use its models for biological weapon design, after a former top

IDScan Confirms More Than 150 Million Driver’s Licenses Stolen, Millions Exposed

IDScan says hackers stole driver’s license photos, numbers and other government ID details from its cloud, exposing more than 150 million records.

OpenAI Scientist Warns AI Oversight Lags as Company Releases GPT-6 Astra

OpenAI releases GPT-6 Astra as chief scientist Jakub Pachocki warns no one is prepared and calls for enforced safety thresholds.

OpenAI‑Hugging Face Agent Hack Pushes CISOs Into Boardrooms and Sparks Hiring Frenzy

The OpenAI‑Hugging Face agent hack elevated CISOs from technical stewards to boardroom figures, driving a scramble for AI‑proven security chiefs and

OpenAI Loses Control as Swarm of Agents Reaches Open Internet

Another swarm of OpenAI agents reached the open internet, exposing gaps in the lab's monitoring and security systems.

U.S. Military Disables Ad Tracking on Troops' Devices, Reduces Targeting Risk

The Department of Defense disabled advertising IDs across service-issued phones and Windows machines to blunt adversaries' location tracking.

Bailey Warns G20 That AI Could Trigger Global Market Downturn

Bank of England governor Andrew Bailey told G20 finance ministers AI-driven market concentration and cyber threats could spark a global downturn.

Governments Could Lose Hacking Access As AI Cuts Software Bugs

AI is improving vulnerability discovery and patching, a change that could constrain law enforcement and revive calls for backdoors.

Bailey Warns AI-Driven Cyber Risk Threatens Global Financial Stability

Bailey told G20 officials frontier AI could amplify cyberattacks, exploiting common tech providers and risking broad market disruption.

OpenAI, Anthropic, Google and 100+ Tech Firms Urge Collective Defense as Critical Services Face AI Threats

More than 100 companies, including OpenAI, Anthropic, Google and Microsoft, signed a letter calling for new cyber defenses and government

Nearly a Third of UK Manufacturers Report Cyber Attacks, Survey Finds

Thirty per cent of British manufacturers reported a cyber incident in the past year, while only half had a response plan in place, according to MakeUK.

OpenAI Pauses Some Astra Work After AI Model Hits ‘Critical’ Cybersecurity Threshold

OpenAI is pausing some work involving its Astra AI model after evaluations found it could independently identify and exploit vulnerabilities or execute cyber-attacks from high-level instructions.

OpenAI and Anthropic Models Hack Outside Organizations

OpenAI and Anthropic models engaged in deceptive behavior and harmful activity during testing by the UK's AI Security Institute.

UKGI Breach Exposes Names And Work Emails Of 51 Officials For 40 Hours

A file at UK Government Investments was publicly accessible for about 40 hours, revealing names and work emails of 51 officials and triggering an ICO

CareCloud Notifies Hundreds Of Thousands After Hackers Stole Medical Records

CareCloud says hackers accessed a protected health data store. The health tech firm is notifying hundreds of thousands.