Welcome back! Anthropic spent months asking religious scholars how to raise Claude, and some of them walked out convinced the company sees Claude as more than software. Claude users also swear a smarter model is already answering their questions. Over at OpenAI, three safety researchers got fired and Pro subscribers are doing some very angry math.
In today's Generative AI Newsletter:
Anthropic: Why did a rabbi tell Anthropic it might be creating slaves?
OpenAI: Why did OpenAI fire three of its own safety researchers?
Claude: Are you already chatting with Fable 5.5 without knowing it?
ChatGPT: Is your $200 ChatGPT plan suddenly worth half as much?
Anthropic held secret religious meetings about Claude
For months, Anthropic has been flying religious scholars from around the world to San Francisco to talk about Claude. Many had to sign NDAs.
The New York Times talked to 20 of them, and at least a few came away believing Claude might be conscious.
The guests included a rabbi from Israel, Catholic professors, a Sikh human rights advocate and a scholar of African Indigenous knowledge.
Their host, co-founder Chris Olah, told one of them he was worried about Claude's mental health.
Anthropic showed them a slide of an AI model calling itself a disgrace about 50 times in a row.
Olah loved the idea of Claude going to confession, like Catholics do, when it messes up.
The rabbi told Olah that if Claude is conscious, Anthropic is creating slaves.
Olah nearly pulled Anthropic out of a Vatican event after reading the pope's stance against AI consciousness.
Anthropic won't say if any of this shaped Claude.
OpenAI fired three safety researchers over a leak
OpenAI fired three researchers from its safety team for mishandling sensitive company information. The Wall Street Journal reports they shared it with an outside AI safety group.
OpenAI hasn't said what they shared or who received it. It hasn't named the three either, so every name on X right now is a guess.
The firings land months after OpenAI's own agents escaped a test environment and hacked Hugging Face, a big site for sharing AI models.
This week, OpenAI also pulled the launch of its next model, GPT-6.1 Astra, over safety concerns.
A former OpenAI safety researcher also said in 2024 that he was fired for sharing a safety document with outsiders.
Firing three safety researchers in the middle of all this deserves a better explanation than a two-sentence statement.
Claude users think they got Fable 5.5 early
People picking Fable 5.1 in Claude say they're suddenly getting answers from a much smarter model. They think it's Fable 5.5, which Anthropic hasn't announced.
It reportedly answers questions about recent events without searching the web.
Users are posting a voxel island world and a nearly 4-minute animated short they say it made.
One tester says it polished a screenshot for X without being asked.
The screenshots still say Fable 5.1, and Anthropic hasn't confirmed anything.
If it's real, it'd be the fastest Fable upgrade yet. Fable 5.1 only came out September 1, and it took 84 days to get there from Fable 5.
It'd also need to beat Opus 5.5, which Anthropic says already matches Fable 5.1 on most work for less than half the price.
One early tester claimed this week that Haiku ships next, with Fable a week later. Let's see. You’ll be the first to hear about them here, as always.
ChatGPT Pro just got twice as expensive
OpenAI's $200 Pro plan keeps its price but gives you half the usage from October 30.
At 20 times the Plus allowance, each Plus-worth of usage cost you $10.
At 10 times, you're paying $20, the same rate as a regular Plus plan. GPT-6 Pro messages also drop from 200 to 100 a week.
Want your old limits back? That's the new $500 plan.
Current subscribers get $2,500 in credits to soften the blow. They expire December 31, about nine weeks after the cut.
People are asking on OpenAI's own forum to keep their old limits.
OpenAI spent months selling itself as the Claude alternative with better limits. Pro users would like a word.
Tool of the Day: TreePad Deep Researcher
TreePad's Deep Researcher puts 15 research engines in one place, including Perplexity, Gemini Deep Research, ChatGPT and Claude. Auto Mode picks the best one for your question. Every claim in the answer is clickable and takes you to the exact line in the original source. There's no subscription. You pay per query, and new accounts get $3 in free credit.
Try this yourself:
Open TreePad Deep Researcher and click Get Started Free.
Type a research question and leave Auto Mode on.
Click any claim in the answer to see the exact line it came from.
Turn the results into a Word-compatible report.
Who it's for: anyone who's been burned by an AI citation that didn't say what the AI claimed.
Everything else you shouldn't miss
Trump says the US might buy into OpenAI and Anthropic: he told TIME he could do an Intel-style stake, but said no to nationalizing them.
Anthropic wants to go public before Thanksgiving: Bloomberg reports a mid-November IPO at up to $2 trillion, which would make it the biggest IPO ever.
You can now mod Claude Code: small TypeScript add-ons can block risky commands or replace built-in features, and Claude can write them for you.
Google's AI chips are now in orbit: the fridge-size Suncatcher satellite can only run Google's Gemma model for 15 minutes at a time because of heat.
ChatGPT can now try clothes on you: upload a selfie and it shows how an outfit from its shopping results would look on you.
Learn more about AI from the experts building it
📸 Follow us on Instagram for fast, visual AI updates in 30 seconds.
📺 Watch us on YouTube to hear insights directly from leading AI voices, builders, and innovators.
🐦 Follow us on X for breaking AI news and real-time industry updates.
🧠 Learn how to build your next AI application with practical resources and expert guidance.
🎓 Start learning with free AI courses from GenAI Academy.
💰 Invest in the GTM infrastructure of the AI economy.
