AI Reality Check — Issue 17

AI Reality Check - Issue 17

Friday 17 July 2026 | Kaye Nicholson | GrowthZone AI | growthzoneai.co.uk

Born analogue. Raised digital. 30 years of real business experience. Now explaining what AI actually means for work.

━━━━━━━━━━━━━━━━━━━━━━

1. Apple Sues OpenAI Over Stolen Hardware Secrets

What happened: Apple filed a lawsuit against OpenAI and two former Apple employees, alleging theft of confidential hardware trade secrets to accelerate OpenAI's push into consumer devices. The complaint alleges a former Apple engineer used an authentication bug to download confidential hardware files, and a former executive emailed himself supplier information and encouraged job candidates to bring Apple parts to interviews. Apple also alleges OpenAI employees approached suppliers using a confidential Apple metal finishing process. More than 400 former Apple staff now work at OpenAI, though Apple argues that alone does not grant access to trade secrets.

Why it matters: this is one of the most direct legal confrontations yet between an established technology giant and a leading AI company. It signals that as AI companies expand into hardware, the competitive and legal boundaries with existing hardware makers are becoming sharply contested.

Who it's for: anyone tracking the competitive dynamics between major tech companies. Businesses evaluating OpenAI's consumer hardware ambitions. Anyone interested in how IP and trade secret law is being tested by AI industry hiring practices.

━━━━━━━━━━━━━━━━━━━━━━

2. GPT-5.6 and Grok 4.5 Both Launched This Week

What happened: OpenAI launched GPT-5.6 as three separate models, Sol, Terra, and Luna, following the preview covered in Issue 16. In the same week, xAI launched Grok 4.5. Separately, Anthropic extended Fable 5 access for a further seven days rather than reverting to standard availability as originally planned.

Why it matters: two major competitors shipped significant new models within days of each other, confirming that the tiered model approach (multiple models at different power and cost levels) is now standard practice across the industry. The Fable 5 extension gives more time for anyone testing Anthropic's most capable model before standard access rules resume.

Who it's for: anyone using ChatGPT, Grok, or Claude for business. Anyone managing AI spend across multiple platforms. Anyone currently testing Fable 5 who now has extra time to do so.

━━━━━━━━━━━━━━━━━━━━━━

3. Meta's Instagram AI Tool Generated Images of People Without Consent by Default

What happened: Meta launched a feature in its Muse Image app that let anyone generate AI images from public Instagram accounts through a simple @ mention. The setting allowing this was switched on automatically, without users' knowledge. People were never notified when someone created an AI image of them, and opting out only stopped future generations, not existing images. Meta pulled the feature days after launch following significant backlash.

Why it matters: this was not a technical bug. It was a design decision, and it shows how AI features involving real people can be built with defaults that favour engagement over consent. For any business using AI tools that touch customer data or images, this is a clear warning about checking actual settings rather than trusting marketing descriptions.

Who it's for: any business using AI tools involving customer images, likeness, or public data. Marketing and social media teams choosing AI tools. Anyone concerned about digital consent and privacy defaults.

━━━━━━━━━━━━━━━━━━━━━━

4. The Federal Reserve Named a Venture Capitalist to Lead Its AI Economic Task Force

What happened: the Federal Reserve named venture capitalist Marc Andreessen to co-lead a new task force studying how AI and other general-purpose technologies affect the economy, alongside economist Charles I. Jones and Xbox CEO Asha Sharma. This is one of five task forces the Fed announced, all expected to report by year end, with findings set to inform Fed policy.

Why it matters: this is one of the clearest signals yet that a major central bank is treating AI's economic and employment impact as a formal, structured policy question rather than background noise. The composition of the task force, including a prominent AI investor, will shape how seriously different perspectives are weighted.

Who it's for: anyone following how AI policy and regulation is developing. Business leaders thinking about the medium-term economic environment for AI adoption. Anyone interested in how central banks are responding to AI.

━━━━━━━━━━━━━━━━━━━━━━

5. OpenAI and Google Sold AI to Companies Blacklisted by the Pentagon

What happened: according to a Financial Times report, Google and OpenAI sold their advanced AI models to Singapore-based subsidiaries of Alibaba, Baidu, and Tencent, companies the US Pentagon has blacklisted over alleged ties to the Chinese military. The transactions are legal because current US regulations do not fully ban Chinese-headquartered firms from using advanced US AI models outside mainland China. OpenAI stated it blocks direct access from mainland China but allows some Chinese-owned businesses to use its services in other jurisdictions.

Why it matters: this highlights a significant gap in current export control frameworks. Legal is not the same as uncontroversial, and this story is likely to accelerate calls for tighter regulation given the ongoing sensitivity around AI and national security.

Who it's for: anyone following AI export controls and national security policy. Businesses concerned about supply chain and platform risk tied to geopolitical tensions. Anyone tracking how AI companies navigate the gap between legal compliance and political risk.

━━━━━━━━━━━━━━━━━━━━━━

Also this week: OpenAI's top safety chief left the company ahead of its expected IPO. An AI actor is now starring in a film. Big Tech has taken on approximately $350 billion in debt to fund AI data centre expansion. Claude Science, Anthropic's AI workbench for scientists, is confirmed to be aiming at medicine discovery.

━━━━━━━━━━━━━━━━━━━━━━

This week in one sentence: Apple sued OpenAI over stolen hardware secrets, GPT-5.6 and Grok 4.5 both launched, Meta's Instagram AI tool generated images without consent by default, the Fed put a task force on AI's economic impact, and OpenAI and Google sold AI to Pentagon-blacklisted Chinese firms.

━━━━━━━━━━━━━━━━━━━━━━

Breaking Update

Three more things landed since this newsletter was first drafted, and they are worth knowing before the week closes.

OpenAI released its first piece of hardware: a $230 mechanical keyboard called kbd-1.0-codex-micro, built with Work Louder, designed specifically to control OpenAI's Codex coding agent. Each key shows live colour states tied to what Codex is doing, whether it is thinking, running, waiting, or finished, and a joystick lets you launch coding jobs directly from the keyboard. The timing is notable. This is OpenAI's first hardware product, released in the same week Apple sued OpenAI over allegedly stolen hardware secrets.

The European Commission ordered Google to open Android to rival AI assistants, using the Digital Markets Act. Under the new rules, people will be able to trigger a competitor's AI assistant using the "Hey Google" wake word, and third-party assistants will be able to carry out tasks inside apps on a user's behalf. Google has opposed the decision strongly, arguing it undermines device-level security protections. This is one of the most direct regulatory interventions yet into how AI assistants are allowed to compete on a dominant platform.

TSMC, the main manufacturer of AI chips worldwide, pledged a further $100 billion investment in Arizona, on top of the $165 billion it had already committed. The pledge came alongside a record quarterly profit, up 77% year on year, and a raised 2026 spending forecast of up to $64 billion, with TSMC citing AI demand it expects to continue through 2030.

Also worth a mention: Mira Murati's startup, Thinking Machines Lab, released its first public AI model, called Inkling. And Apple is reportedly in talks to acquire an AI chip startup after its own next-generation AI server chip was delayed past its planned 2026 launch, which adds useful context to Apple's broader position in this week's lawsuit against OpenAI.

Who it's for: anyone tracking competitive and regulatory developments in AI hardware and platform access. Businesses using Android-based AI assistants or considering third-party AI tools that depend on platform access. Anyone following the Apple and OpenAI dispute, since both of today's hardware stories add useful context to it.

Born analogue. Raised digital. 30 years of real business experience explaining what AI actually means for work.

— Kaye Nicholson | GrowthZone AI | growthzoneai.co.uk

Subscribe at growthzoneai.co.uk | Follow @GrowthZoneAI

#AIRealityCheck #GrowthZoneAI #UKBusiness #NorthEast