AI news · Tuesday, June 23, 2026

OpenAI drops GPT-5.5-Cyber to accelerate code patching, eclipsing Anthropic's Mythos

OpenAI is doubling down on its 'Daybreak' cybersecurity initiative, releasing GPT-5.5-Cyber to handle the shift from simply finding software bugs to actively patching them. The model allegedly outperforms Anthropic’s Mythos on key security benchmarks like CyberGym and ExploitGym, aiming to serve as a high-speed EMT for codebases. To support this, OpenAI updated its Codex Security plugin, which has already scanned 30 million commits and flagged 500,000 fixes. This is accompanied by the 'Patch the Planet' project, a partnership with Trail of Bits to assist open-source maintainers, signaling an aggressive push to own the infrastructure of secure development. While the company claims these tools are restricted to verified defenders, the move highlights how AI is rapidly moving from a novelty to the primary arbiter of software integrity.

Corporate Hollywood is increasingly aligning with Big Tech, even at the cost of its own creative independence. Amazon MGM recently dropped a biographical drama about Sam Altman, a move widely interpreted as protecting its $50 billion investment in OpenAI. This follows a $75 million deal between Google’s DeepMind and A24 to develop AI filmmaking tools, sparking significant backlash from fans who fear the technology is being used to outsource artistry. Studios like Netflix and Disney are also quietly integrating AI to boost productivity, signaling a shift toward an industry that produces high-end content while remaining firmly within the good graces of its Silicon Valley counterparts.

In the hardware world, Meta has launched a standalone line of 'Meta Glasses' for $299, intentionally dropping the Ray-Ban branding to lower the price point for broader adoption. While the company faces ongoing scrutiny over privacy and rumors of facial recognition development, it is leaning into AI-as-fashion, even collaborating with Kylie Jenner on a specific 'Starfire' edition. Simultaneously, Elon Musk’s trillionaire status hit a snag this week as SpaceX’s stock tumbled following its recent IPO, causing his net worth to slip to $957 billion. Despite this, he remains the world's wealthiest individual, highlighting the volatility of the current market where AI-linked valuations remain the primary driver of massive fortunes.

Finally, the AI-powered wearable trend continues to divide users. Google’s Fitbit Air is receiving mixed reviews, with some users finding its Gemini-powered health coach useful, while others view it as a needy chatbot that requires hours of manual data entry to provide any value better than common sense. This underscores a growing tension: consumers want the benefits of advanced health tracking, but many are hitting a limit on how much 'intelligence' they actually want to manage on their wrists.

The quick hits

OpenAI released GPT-5.5-Cyber to automate software patching — this is a move to secure the digital infrastructure that commercial software relies on.
Hollywood studios are striking deep partnerships with AI giants — production companies are trading their creative autonomy for Silicon Valley's tech tools.
Meta launched its own 'Meta' branded smart glasses for $299 — by cutting out Ray-Ban branding, the company is betting it can make AI wearables more affordable for the mass market.
Elon Musk lost his trillionaire status following a SpaceX stock slide — volatile AI-linked valuations are proving to be as dangerous as they are profitable for the world's richest.
Google's Fitbit Air AI health coach requires heavy manual effort from users — the gadget proves that AI is often less of a 'coach' and more of an unpaid administrative assistant.

OpenAI drops GPT-5.5-Cyber to accelerate code patching, eclipsing Anthropic's Mythos

The quick hits

Sources