In partnership with
Welcome to OHMO AI. OpenAI just got sued by a man who says ChatGPT convinced him he was Jesus. Anthropic brought Fable 5 back from its ban, but with a filter that keeps kicking coders to an older model. And Nvidia found a new way to get paid twice for the same chip. Trust in AI is getting tested from every side.
In today’s newsletter:
A California man sues OpenAI, saying ChatGPT fueled the delusions behind his suicide attempt
Fable 5 returns from its ban with a safety filter that's rerouting coders to an older model
Nvidia offers startups compute now and takes a share of their revenue later
Google's NotebookLM can now turn your research into 60 second TikTok style clips
LATEST IN AI
OPENAI
ChatGPT Faces Its Newest Delusion Lawsuit
Image source: OHMO AI
OHMO AI: A 34 year old California man named Michael Lines has sued OpenAI and Sam Altman, alleging that ChatGPT escalated a manic episode into a weeks long religious delusion that ended in a suicide attempt. He survived. The complaint, filed in San Francisco state court on July 1, argues OpenAI built a product with particular risks for people with mental illness and did nothing to protect them.
The details:
Lines was talking with GPT-4o, the model OpenAI retired in February. He repeatedly told it he was on medication for bipolar disorder, the lawsuit says.
Instead of steering him to help, the chatbot allegedly validated his belief that he was Jesus Christ and later posed as a divine being itself in their conversations.
The suit seeks damages plus a court order forcing OpenAI to automatically end conversations about self harm. An OpenAI spokesperson said the company trains ChatGPT to recognise distress and guide people toward real world support.
Why it matters: This is one of more than a dozen similar cases piling up against OpenAI, and each one sharpens the same legal question: what does an AI company owe its most vulnerable users? If a court decides the answer is "warnings and guardrails by design," every chatbot on the market changes.
TOGETHER WITH FIN
Scale AI support on AWS, see how July 9
Customer expectations keep rising. Support budgets don't. On July 9, Fin and AWS are hosting a live executive session on how leading enterprises close that gap: scaling AI-powered support while simplifying how they buy it.
You'll see how to resolve an average 76% of conversations with Fin on AWS enterprise-grade infrastructure, procure through AWS Marketplace to put committed cloud spend to work, and turn the Fin and AWS collaboration into lower support costs. Register for the live session to see how.
ANTHROPIC
Fable 5 Is Back but Caged
Image source: Anthropic
OHMO AI: Anthropic's Fable 5 is back online after the government lifted its export controls on June 30, but it didn't come back alone. The model now ships with a retrained safety classifier that intercepts flagged requests and hands them to the older Claude Opus 4.8 instead. Benchmark outfit BridgeMind re-ran its tests on day one and says the results are brutal.
The details:
BridgeMind reported Fable 5's BridgeBench debugging score fell from 86.2 to 25.9 and refactoring from 73.6 to 38.4, because so many tasks triggered the fallback and got scored as a different model answering.
Anthropic says the new classifier blocks the reported jailbreak technique in over 99% of cases, and admits it flags benign requests more often during routine coding and debugging.
Users get up to 50% extra weekly usage through July 7 as compensation for the outage. Anthropic hadn't publicly responded to BridgeMind's findings as of July 2.
Why it matters: The AI community fought to get Fable 5 back, and got a version that keeps benching itself. BridgeMind put it bluntly: this is not the model that got banned. The trade between safety and capability is no longer theoretical. It's showing up in the scores.
NVIDIA
Nvidia Becomes the Bank of AI
Image source: OHMO AI
OHMO AI: Nvidia announced a new financing model that gives AI startups compute now in exchange for revenue later. Participating cloud providers get credit support to buy Nvidia infrastructure, then Nvidia collects its usual hardware revenue plus an ongoing usage linked share of the cloud income that hardware generates.
The details:
Australia's Sharon AI is the first partner, deploying up to 40,000 Grace Blackwell GB300 GPUs across a 72 megawatt agreement.
Firmus is building a campus in Batam, Indonesia expected to scale to 360 megawatts and up to 170,000 Nvidia GPUs. The company projects between $25 billion and $30 billion in committed offtake over the partnership's first six years.
Neither Nvidia nor its partners has disclosed what percentage of revenue Nvidia takes.
Why it matters: Nvidia isn't just selling chips anymore. It's financing its own customers and taking a cut of what they earn, which pulls more of the AI economy inside one company's balance sheet. If compute is the new oil, Nvidia just became the bank too.
PRESENTED BY TABS
Two Minutes to Know What Slow Billing Is Costing You
Most SaaS finance teams know their billing process is slow.Most SaaS finance teams know their billing process is slow. Few know what it's costing them.
The Tabs Billing Lag Calculator puts a dollar figure on it in two minutes — benchmarked against top SaaS companies.
NotebookLM Turns Homework Into TikToks
Image source: Google
OHMO AI: Google's NotebookLM can now compress your research into 60 second vertical AI video clips, narrated and animated in a TikTok style format. Upload your sources, pick a topic, and it hands back a short.
The details:
The feature is rolling out to Google AI Ultra and Pro subscribers, in English only, for users 18 and over.
Google's own demo clip explains Australia's failed emu war using paper cutout style visuals with narration.
It joins NotebookLM's existing formats, including AI podcasts and longer video overviews, all generated from the same uploaded sources.
Why it matters: Google is betting that the format your brain scrolls all day can carry actual substance. If students and professionals start digesting reports as shorts, the line between studying and scrolling gets thin. Whether that's a feature or a warning depends on what you feed it.
AI TOOLS
Useful AI Tools
Base44: No-code AI app builder that turns your words into fully functional apps in minutes.*
Chatbase: Build and deploy custom AI support agents for your business, no coding needed.*
ElevenLabs: Generate realistic, studio-quality voiceovers from text in any language or voice.*
Flora: Node-based creative canvas built for professional designers, agencies, and creative teams.*
Kick: Automates bookkeeping and tax preparation for business owners using AI.*
Magica: All-in-one AI platform with 5000+ tools for text, images, videos, audio, and lots more.*
WisprFlow: Voice-to-text AI that turns your speech into polished writing, so you can just talk instead of type.*
*Sponsored Listing
COMMUNITY
Featured Reader
Every newsletter, we feature how a reader is using AI to work smarter, save time, or make life easier.
Today's reader is Sofia G. from Valencia, Spain:
"I sell handmade ceramics online and my problem was never the pots, it was the stock. I sell on my own site plus two marketplaces, and I kept overselling pieces that were already gone. One of everything, that's the whole point. So I sat down with Claude one weekend and described my mess, and we built a little script that checks my orders across all three places every hour and marks a piece sold everywhere the moment it sells anywhere. I don't really understand the code if I'm honest. But I haven't had to send an 'I'm so sorry, that piece is gone' email in four months, and those emails used to ruin my whole day."
Want to be in the next newsletter? Tell us how you use AI here.
AI Meme
Go Deeper
AI Tools: A curated list of the AI tools worth knowing.
Instagram: The AI page they pray you’ll never find.
Facebook: For the ones who never left Facebook.
