Apple and OpenAI Go to War
The rivalry between Silicon Valley's biggest names turned openly hostile. On Friday, July 10, Apple filed suit against OpenAI in federal court in San Jose, accusing the ChatGPT maker of a "coordinated campaign" to steal trade secrets in order to build competing AI-powered hardware devices, according to Reuters. It is a striking escalation: the world's most valuable consumer-hardware company is now fighting the most prominent AI lab in court over the future of the device on your desk — or in your hand.
The timing is pointed, because OpenAI is having one of its busiest stretches in months.
OpenAI Ships, Kills, and Reshuffles
A day before the lawsuit, on July 9, OpenAI publicly released GPT-5.6, its most powerful model family yet, according to Reuters and Axios. Alongside it came ChatGPT Work, an AI agent built to carry out entire jobs rather than merely answer questions — a clear signal of where the company thinks its models are headed.
OpenAI also expanded into voice with GPT-Live, a new generation of models now powering ChatGPT Voice. Their "full-duplex" architecture lets them listen and speak at the same time, edging closer to natural conversation.
But not everything survived the week. OpenAI is retiring Atlas, the standalone web browser it built to challenge Chrome, and folding its capabilities into ChatGPT, per Mashable. The company frames it less as a retreat from browsers than as a change of strategy — the browser becomes a feature, not a product.
The momentum was tempered by a personnel loss. Fidji Simo, one of OpenAI's most senior executives, said on Thursday, July 9 that she will step down from her full-time role to focus on her health amid a chronic illness, moving to a part-time capacity, according to Reuters.
Meta's Rough Week: Muse Rolled Out, Then Rolled Back
Meta spent the week both advancing and retreating. It launched Muse Image, a feature letting anyone generate AI images based on any public Instagram account, ZDNet reported. The backlash was swift, and within days Meta reversed course, announcing it would discontinue the feature after sustained public criticism.
On the developer front, Meta Superintelligence Labs released Muse Spark 1.1 to developers on July 9, positioning the model directly against OpenAI and Anthropic. The company is clearly determined to compete at the frontier — even as its consumer-facing AI stumbles on trust and privacy.
Meta also lost control of an asset. According to Tom's Hardware, Beijing is forcing Meta to give up the AI startup Manus, with Tencent reportedly in talks alongside other investors to raise the roughly $2 billion needed to take it on. It is a reminder that AI's biggest deals now answer to governments as much as to boardrooms.
Anthropic Peers Inside the Machine
Amid the corporate drama, Anthropic offered a glimpse of the research frontier. The company says it has found a way to look inside its Claude models' internal processing — what it loosely describes as reading their "thoughts," per Tom's Hardware. If it holds up, the work speaks to one of AI's deepest problems: understanding why these systems do what they do.
The Question Underneath It All
Running beneath the week's launches and lawsuits is a quieter theme. AI has moved from novelty to workplace fixture, but a cluster of recent coverage suggests the pressing question is no longer whether to adopt it — it's whether companies can actually rely on it to work. GPT-5.6 and ChatGPT Work promise agents that do the job. Meta's Muse reversal shows how fast trust can erode. Between the ambition and the anxiety sits the story of 2026's AI industry.