Apple Details OpenAI Hardware Theft Scheme.

Mazen Al Sabbagh

— 4 min read
Share

TL;DR

  • Apple v. OpenAI: Apple initiated federal litigation today, alleging systemic trade secret misappropriation by OpenAI, implicating Jony Ive's io Products in the scheme.
  • SK Hynix Surge: The Korean memory leader secured $26.5B in the largest foreign US listing to date, closing up 13% with a $1.27T valuation.
  • UAE Chip Access: The Commerce Department reclassified the UAE to Country Group A:5, enabling license-free advanced AI chip exports to approved entities.
  • Frontier Model Debuts: GPT-5.6 and Grok 4.5 launched publicly; early assessments place Grok fourth on intelligence benchmarks, noting a 54% hallucination rate.
  • Copyright Sanctions: A consortium of newspapers, led by The New York Times, moved for sanctions against OpenAI, citing alleged misrepresentation regarding content search capabilities.

Lead Story: Apple Details OpenAI Hardware Theft Scheme.

Apple filed suit today in the Northern District of California, asserting that OpenAI orchestrated extensive trade secret theft, impacting "every level, from members of its Technical Staff to its Chief Hardware Officer," all in pursuit of its forthcoming consumer hardware initiatives (CNBC, TechCrunch).

Key allegations detail actions by Tang Tan, OpenAI’s hardware chief and a former Apple VP, who purportedly instructed active Apple employees to present "actual parts" in interviews for "show and tell" sessions, leveraging Apple's internal codenames to extract confidential data (Fortune). Another former employee, Chang Liu, is accused of departing with an Apple laptop, while OpenAI allegedly coached exiting staff on evading Apple’s security protocols. Jony Ive's io Products, acquired by OpenAI for $6.4B, is explicitly named in the illicit scheme (CNN).

This litigation marks a significant pivot from the 2024 partnership that integrated ChatGPT into the iPhone. Over 400 former Apple employees now work at OpenAI, and Apple has confirmed its rebuilt Siri, scheduled for an autumn release, will operate on Google Gemini instead (Axios). The timing is critical: OpenAI's public market ambitions, projecting valuations up to $1T, now face injunction threats against the core hardware narrative underwriting that valuation.

In Other News

SK Hynix executes the largest foreign US listing ever. The HBM sector leader raised $26.5B on Nasdaq via 177.9 million ADRs at $149, oversubscribed sevenfold. This surpasses Alibaba's 2014 record for a foreign debut and trails only SpaceX’s $86B IPO last month among all US share sales. Shares advanced as much as 17% intraday before settling nearly 13% higher, establishing a $1.27T market capitalization under ticker SKHYV (Forbes, Yahoo Finance). Chairman Chey Tae-won’s candid assessment – "All my customers said that’s not enough" – underscores the robust AI infrastructure demand driving this valuation, setting a precedent for imminent OpenAI and Anthropic IPOs.

US opens license-free AI chip exports to the UAE. The Commerce Department’s Bureau of Industry and Security reclassified the UAE from Country Groups D:3/D:4 to A:5, effective today, positioning the first Arab nation among America's primary technology partners. Approved entities, including G42 and Core42, plus US firms operating locally, can now receive advanced AI chips and servers without individual licenses (BIS). Senator Elizabeth Warren criticized a provision impacting Emirati fund MGX as "corrupt" (CNBC). The market awaits whether Saudi Arabia will demand reciprocal treatment.

GPT-5.6 and Grok 4.5 face their initial market verdicts. OpenAI unveiled its Sol, Terra, and Luna models across ChatGPT, Codex, and the API, with Sol priced at $5/$30 per million tokens. Concurrently, ChatGPT Work debuted, integrating Codex as an agent with 15 @mention-activated integrations (OpenAI, Testing Catalog). SpaceXAI's Grok 4.5 launched the same morning at a disruptive $2/$6 price point. Independent analysis today shows Grok 4.5 ranking fourth on the Artificial Analysis Intelligence Index (54), behind Claude Fable 5, GPT-5.5, and Claude Opus 4.8. While exhibiting strong token efficiency, its hallucination rate escalated to 54% from Grok 4.3's 25% (Build Fast with AI).

Newspapers ask court to sanction OpenAI in copyright fight. A consortium of news organizations, including The New York Times and Daily News, informed a federal judge Thursday that OpenAI falsely asserted an inability to search its models and output logs for copyrighted content, despite having conducted "multiple searches" pre-litigation, and deleted billions of pertinent ChatGPT conversations (TechCrunch, Washington Post). Coupled with the Apple lawsuit, OpenAI now confronts allegations of improper discovery conduct on two fronts during its critical IPO preparatory phase.

X / Social Pulse

Grok 4.5 dominated launch-day discourse. Elon Musk claimed it "landed at number 1 on multiple benchmarks," while a Hacker News thread on political bias splintered between concerns over Musk's influence and counter-reports suggesting Grok is "more politically correct than GPT and Gemini." The 54% hallucination statistic widely circulated as a counterpoint to benchmark victories. Developers testing GPT-5.6 also noted that the economical Luna tier occasionally outperforms the mid-tier Terra on routine coding tasks—an awkward outcome for OpenAI's tiered pricing strategy.

One to Watch

Gemini 3.5 Pro's GA date. Google's flagship model is now six weeks past its June target, with an unconfirmed July 17 date circulating. Google positions itself as a quiet beneficiary this week; the Apple-OpenAI rupture confirms Siri's autumn launch on Gemini, securing the iPhone distribution channel OpenAI forfeited.

Quick Hits

  • Fed AI Task Force: Fed Chair Kevin Warsh appointed Marc Andreessen as co-lead of a new task force on AI's economic impact, alongside Stanford's Charles I. Jones and Microsoft's Asha Sharma; recommendations are expected by year-end (Washington Post).
  • Illinois Frontier AI Law: Illinois enacted SB 315, becoming the first state to mandate frontier-model safety, requiring transparency, catastrophic-risk assessments, and whistleblower protections, with support from OpenAI and Anthropic (Transparency Coalition).
  • UN AI Governance Warning: The inaugural UN all-nations AI governance dialogue concluded in Geneva with its 40-member scientific panel stating that science "cannot guarantee" advanced AI will prevent catastrophic harm.
  • China Companion AI Rules: China's Interim Measures for Anthropomorphic AI Interaction Services, governing companion AI, become effective July 15.
  • Anthropic Cowork Mobile: Anthropic released Claude Cowork mobile applications for iOS and Android, enabling users to delegate and monitor persistent agent sessions remotely.

This Friday delivered a rupture in a pivotal tech partnership, validated the AI-infra IPO pipeline with a $26.5B debut, and subjected two frontier models to initial independent scrutiny. The overarching theme is the capital market's direct influence: each development today either bolsters or jeopardizes an impending AI listing.

Sources

Apple v. OpenAI: CNBC · TechCrunch · Fortune · CNN · Axios SK Hynix: Forbes · Yahoo Finance · CNBC UAE exports: BIS · CNBC Models: OpenAI · Testing Catalog · Build Fast with AI OpenAI copyright case: TechCrunch · Washington Post · Al Jazeera Quick hits: Washington Post · Axios · Transparency Coalition

Lock in. M. mazen@thorterminal.com

Read more