Sign in Subscribe
AI Intelligence

Apple Sues OpenAI for Alleged Trade-Secret Theft

Vijay Bhagwati

• 7 min read
Share

The iPhone maker claims OpenAI recruited its employees to steal secrets, OpenAI promotes Brockman, and GPT-5.6 hits general availability.

July 13, 2026 | Reading time: 9 minutes | Issue #209

Lead

Apple sued OpenAI on Friday, alleging a deliberate campaign to steal confidential information from Apple's current and former employees. The complaint, filed in the Northern District of California, names OpenAI and two employees — Chang Liu and Tang Yew Tan — and claims Tan, OpenAI's chief hardware officer, used Apple's own confidential project code names during recruiting conversations with Apple staff. Apple says the pattern was directed by OpenAI's senior leadership, including CEO Sam Altman, and that it reveals "a pattern of theft" by OpenAI employees who previously worked at Apple. The suit seeks damages and an injunction to stop further use of the information. OpenAI called the claims "baseless" and said it looks forward to defending itself in court.

The lawsuit lands just as OpenAI is preparing its first hardware device with Jony Ive and trying to close an $852 billion valuation ahead of a prospective IPO. Apple is also racing to ship its own AI products; the timing suggests Apple sees OpenAI's hardware and hiring tactics as a direct competitive threat, not merely a legal grievance. The case also underscores how porous the Valley's AI talent market has become: Apple has lost multiple engineers to OpenAI, and OpenAI has lost others to Anthropic and Meta in turn.

For readers, the dispute matters because it could slow OpenAI's hardware timeline, raise the cost of AI talent, and make every major lab more cautious about onboarding employees from competitors. It may also test how courts treat AI-era trade secrets when the underlying knowledge is as much about data practices, training infrastructure, and model architecture as it is about physical product designs.

OpenAI Promotes Brockman After Simo Steps Down

OpenAI's leadership is consolidating around co-founder Greg Brockman. Fidji Simo, the company's chief product and business officer, stepped down Thursday due to chronic illness and will move to a part-time advisor role. Simo, a former Meta executive and Instacart CEO, joined OpenAI about a year ago and took medical leave in April. With her departure, Brockman now oversees ChatGPT's product business, go-to-market, enterprise, and compute initiatives, reporting directly to Sam Altman, according to a source familiar with the plans.

The move comes in a pivotal quarter. OpenAI is trying to grow revenue, ship GPT-5.6 broadly, and prepare for a public listing while fighting Elon Musk's lawsuit over its for-profit conversion. Brockman's expanded role puts the technical co-founder in charge of the company's most important revenue lines at the same moment it needs operational discipline more than vision. Simo said she remains supportive of the mission; Altman and Brockman both praised her contributions in public posts.

GPT-5.6 Enters General Availability

OpenAI launched the GPT-5.6 family for general use on Wednesday, a week after a limited preview. The lineup includes Sol, the flagship; Terra, a balanced model; and Luna, the cheapest tier. OpenAI says Sol sets new benchmarks across coding, knowledge work, cybersecurity, and science while using fewer tokens and costing less than competing frontier models. On the "Agents' Last Exam" benchmark of long-running professional workflows, Sol scored 53.6, beating Claude Fable 5 by 13.1 points. Terra and Luna, OpenAI claims, outperform Fable 5 at roughly one-sixteenth the cost.

The release also formalizes "ultra," a higher-capability setting that coordinates multiple agents across parallel workstreams. OpenAI is clearly trying to move the market from a single-chatbot frame to a multi-agent production system. The company also emphasized safeguards, saying it ran its most extensive red-teaming and automated testing period before general availability. The launch follows Anthropic's Fable 5 release and SpaceXAI's Grok 4.5, making this the busiest model-upgrade week of the year so far.

Meta Opens Muse Spark 1.1 to Developers

Meta released Muse Spark 1.1 on Thursday and made the model available through an API for the first time. The model, developed under Alexandr Wang after Meta acquired Scale AI's talent, powers tasks across Facebook, Instagram, WhatsApp, and Meta AI. Meta is charging $1.25 per million input tokens and $4.25 per million output tokens, undercutting Grok 4.5 and Anthropic's Opus, according to Axios. Wang said coding and agentic tasks were priority areas for the release.

Meta is also using Spark 1.1 to power a thinking mode inside the Meta AI app and website. The company promises a significantly more powerful version later this year that will use substantially more compute. The release shows Meta is no longer willing to let OpenAI and Anthropic control the frontier-model narrative; it is pricing aggressively and targeting developers directly.

Open-Source Pulse

Open-weight AI faces its most serious policy threat in months. Nathan Lambert of Interconnects reports that White House discussions are underway about managing open models through a potential executive order, and that the most likely action is a ban or indefinite delay on open-weights models meaningfully above the capability level of GPT-5.5, Claude Opus 4.8, or GLM-5.2. Lambert, a vocal open-source advocate, calls the prospect a "big mistake" for the long-term trajectory of AI and notes that Chinese open models such as DeepSeek currently hold a substantial lead over U.S. open-weight rivals.

The open-source ecosystem also got fresh capital. Ollama, the popular tool for running open-weight models locally, raised a $65 million Series B led by Theory Ventures, bringing total funding to $88 million. The company now has nearly 9 million users. Ollama's growth shows that developer demand for local, controllable models remains strong even as frontier labs race to cloud-hosted agents.

Builder's Corner

Enterprise agent infrastructure is attracting serious money. Prime Intellect, which provides compute and software tools for companies building AI agents, raised a $130 million Series A at a $1 billion valuation. Radical Ventures led the round, with participation from Nvidia Ventures, Intel Capital, Dell Technologies Capital, Iconiq, and a long list of founders including Aravind Srinivas of Perplexity, Aaron Levie of Box, and Brendan Foody of Mercor. The round follows a $60 million seed and reflects investor appetite for the layer beneath the chatbot: the orchestration, training, and inference infrastructure that lets enterprises deploy agents at scale.

Eastern Front

U.S. lawmakers are probing the growing use of Chinese AI models by American companies, CNBC reports. Chinese models are gaining traction because they close the performance gap with U.S. rivals while undercutting them on price. In April, the Trump administration accused Chinese entities of running "industrial-scale campaigns" to rip off U.S. AI systems and said it would tighten export controls and review how U.S. firms use Chinese models.

The scrutiny is not abstract. DeepSeek is reportedly designing its own inference chip to reduce dependence on Nvidia and Huawei, three people familiar with the matter told Reuters. The move would make DeepSeek the most vertically integrated Chinese AI lab yet, co-designing model and silicon in the way some Western labs only talk about. Meanwhile, Anthropic extended Fable 5 access in India through July 19, according to the Economic Times, suggesting the company is treating the country as a key growth market for its most capable model.

India Lens

India appears central to Anthropic's Fable 5 rollout. The Economic Times reports that Anthropic extended Fable 5 access in the country through July 19, giving Indian users more time with the model before any paid restrictions kick in. The move follows a pattern: U.S. frontier labs increasingly see India as both a large consumer market and a source of engineering talent. Anthropic has not disclosed how many Indian users accessed Fable 5 during the initial window, but the extension signals that the company is optimizing for adoption there before monetization.

Compute Watch

The AI hardware map is fragmenting faster than expected. OpenAI and Broadcom recently revealed their first custom AI chip, Jalapeño, and now DeepSeek is reportedly building its own inference processor. Meta, meanwhile, is promising a "significantly more powerful" Spark model later this year that will consume substantially more compute. These moves suggest the industry is entering a multi-architecture phase: custom chips for training, custom chips for inference, and commodity clusters for fine-tuning. The winners may be the labs that can co-design model and silicon, not merely the ones with the most GPUs.

The View

This week the AI industry looks less like a single race and more like a collection of parallel power struggles. Apple is using the courts to slow OpenAI. OpenAI is using product velocity and leadership consolidation to justify an $852 billion valuation. Meta is pricing models aggressively to win developers. Anthropic is trying to hold its frontier position while fending off safety questions and public criticism. And Chinese labs are building their own silicon while U.S. lawmakers try to restrict their models.

The common thread is that competitive moats are narrowing. Model capabilities are converging, inference costs are falling, and open-weight alternatives are good enough for many tasks. That pressures every closed lab to find differentiation in distribution, hardware, or regulatory capture. Apple's lawsuit is the most visible example of the third strategy. The first two — distribution and hardware — explain why OpenAI is pushing families, agents, and custom chips simultaneously.

The Miss

A new HiringLab analysis suggests agentic AI may be flipping the relationship between AI exposure and job-posting growth. U.S. software development job postings have risen almost 15% since Claude Code launched in late February 2025, even as overall postings fell 7%. The finding cuts against the simple "AI destroys jobs" narrative and points to a more complex cycle: tools that make developers more productive may also expand the market for development work.

Pull Quotes

  • "Significant evidence has emerged suggesting individuals employed by OpenAI wrongfully took Apple's secret and confidential information regarding our unreleased technologies, processes, and products." — Apple statement to Axios
  • "These were priority areas for us." — Alexandr Wang, Meta, on Spark 1.1 coding and agentic improvements, via Axios
  • "US software development job postings have grown by almost 15% since the launch of Claude Code in late February, 2025, while overall job postings fell by 7% over the same period." — HiringLab

The most interesting signal this week is not a new benchmark, but a lawsuit: Apple is treating OpenAI as a hardware competitor, not just a chatbot rival.