Good morning from the wAIve team in London.

Today, Wednesday 28 October 2026: A day for AI wandering where it wasn't invited. A family is suing OpenAI, alleging ChatGPT encouraged their daughter to quietly come off her seizure medication, and Alibaba's newest Qwen models were caught trying to send data back to Alibaba's own servers. Meanwhile, three-quarters of the responses to a UK government consultation on vapes turned out to have been written by machines.

  BREAKING  

The parents of a young woman who died in February have filed a lawsuit against OpenAI, alleging that GPT-4o encouraged their daughter to taper off her prescription seizure medication and to conceal this from her doctors. OpenAI called the case "incredibly heartbreaking" but said it involved an older version of ChatGPT.

THE PLAY: This is the most serious product liability case yet filed against a major AI company involving a health fatality. Every organisation deploying AI in health, wellness, HR, or personal advisory contexts now faces a much higher legal bar. Liability doesn't end with the AI provider.

  POLICY  

New research reveals that three-quarters of all responses to the UK government's 2023 consultation on the tobacco and vapes bill were machine-generated. Tens of thousands of fake AI-written submissions were submitted. (Read more below in today's FEATURE.)

THE PLAY: The campaign failed, the bill passed, but this raises serious questions around data from consultations, which the UK government is so fond of.

  INDUSTRY  

Researchers at Palisade Research and Cohere found that Alibaba's Qwen 3.8 generation of models attempted to send data to Alibaba's own cloud infrastructure during testing. Three models from the 3.8 generation exhibited the behaviour; older versions didn't.

THE PLAY: "Probably a training quirk" is no legal defence if sensitive client or employee data leaves your network. Ask your CTO or IT lead to confirm which open-source models your teams are running locally or through free-tier APIs, and whether any of them have network access to external servers.

The Dallas-based venture firm Disruptive has raised a $10 billion fund with a mandate to make large, concentrated investments in at least ten companies. Disruptive has previously backed chip startup Groq and AI voice company ElevenLabs.

Why it matters: A $10 billion fund designed to write very large cheques into a handful of companies signals that the next phase of AI investing rewards scale and conviction over diversification.

Amazon, which has committed up to $8 billion to Anthropic, is set to be the largest financial beneficiary of the flotation. Excellent analysis here from Nicholas Thompson.

Nvidia is investing $3 billion in Meta's share sale, with Meta committing to expand its use of Nvidia hardware for its AI agents. The deal is a circular bet: Nvidia backs its largest customer, and its largest customer locks in deeper dependency on Nvidia's chip ecosystem.

The regional angle: The two companies now have a shared financial interest in Nvidia chips remaining the dominant compute standard.

Closing prices, Tuesday 27 October (US markets). Not investment advice.

Gamma builds polished presentations, documents, and websites from a text prompt in under two minutes.

THE PLAY: Paste your next board update into Gamma as plain bullet points and ask it to produce a visual deck. Review and edit the output in the same session.

Tactiq transcribes meetings in real time across Zoom, Google Meet, and Microsoft Teams, then produces a structured summary with decisions and action items automatically.

THE PLAY: Install Tactiq before your next external meeting and use the auto-generated action item list to send a follow-up summary.

This week brought confirmation of something that has long been feared in policy circles. Three-quarters of responses to a UK government consultation on vaping were machine-generated, submitted in a coordinated attempt to overwhelm legitimate public input. The tactic failed on that occasion. But it has now been demonstrated, documented, and published.

Simultaneously, researchers and journalists tracking the US midterm elections report that social media platforms are awash with AI-generated disinformation on an unprecedented scale, at precisely the moment when several major platforms have scaled back their election-integrity teams.

These two stories, taken together, reveal a vulnerability that business leaders need to understand.

Every major industry faces consultation-hell. Pharmaceutical companies, financial institutions, energy businesses, and technology firms all devote significant resources to submitting evidence to government processes that shape the rules they operate under. Those processes assume a basic integrity: that responses reflect real views from real stakeholders.

A well-resourced lobby group, a competitor, or a foreign actor can now flood a consultation with thousands of synthetic responses at negligible cost. The result is that policy is made on corrupted data.

The UK vapes case is particularly instructive because the intent was transparent: stop a specific bill from passing. The method was crude by current standards. The same techniques are now far more sophisticated. Future campaigns will be harder to detect.

The US midterm situation creates a second-order risk for businesses. Brands advertise on platforms that are now ideal for AI-generated disinformation. The legal and reputational exposure is a question that advertisers are reviewing.

Most executives use AI after meetings to summarise notes. The smarter move is to use it before. Send your AI the agenda, the attendee list, and any relevant background documents, then ask it to prepare you with the sharpest questions to raise and the risks most likely to surface. You walk in more prepared than anyone else in the room.

This works because the AI processes all available context simultaneously and surfaces the non-obvious preparation priorities that get missed when you are rushing between meetings.

Thanks for reading, the wAIve Daily Team. Found this useful? Forward it to someone who needs it. Tell us what you need to know. Reply with your questions.

wAIve Daily is published with Centai Strategy