Cybersecurity researchers from Calif used the widely discussed AI model Claude Mythos to bypass Apple’s Memory Integrity Enforcement, or MIE, protection mechanism in macOS.
By Daniel Mercer The Pentagon is pressing leading AI companies — including OpenAI, Anthropic, Google, and xAI — to make their AI tools available on classified military networks without the usual usage restrictions, according to Reuters, citing multiple sources
By Daniel Mercer OpenAI is deploying a specialized version of ChatGPT to detect internal information leaks, according to The Information, citing a source familiar with the matter. When a news article about internal developments at OpenAI is published, security staff reportedly feed the text into this dedicated ChatGPT system, which has access to internal documents as well as employees’ Slack and email communications
By Alex Rowland Former OpenAI researcher Zoë Hitzig has resigned over the introduction of advertising in ChatGPT. In a commentary for The New York Times, she made clear that she no longer trusts her former employer’s direction.
By Chris Borden OpenAI has deployed a specialized version of ChatGPT on GenAI.mil, a secure platform operated by the U.S. Department of Defense.
By Chris Borden Gemini-based Google Translate can be hacked with words. Users can exploit so-called prompt injection—embedded instructions in natural language—to directly address the underlying language model instead of receiving a translation.
By Chris Borden OpenAI is in talks with the Abu Dhabi–based company G42 about developing a custom version of ChatGPT for the United Arab Emirates, according to Semafor. The tailored version would support local Arabic dialects, reflect local political perspectives, and include specific content restrictions—initially for use by the UAE government.
By Chris Borden Anthropic’s safety training fails when Claude operates a graphical user interface
By Alex Rowland The popular open-source AI agent OpenClaw can be fully compromised through maliciously crafted documents, allowing attackers to install a persistent backdoor and take control of a user’s computer, security researchers have shown.
By Chris Borden OpenAI has accused Elon Musk’s AI company, xAI, of systematically and deliberately destroying evidence in an ongoing court case.
ES 
EN