Welcome, AI Enthusiasts!
Over dinner, Sam Altman talked trillions for compute, fixing GPT-5, and even eyeing Chrome if regulators pry it loose. Also on your radar, Claude now walks away from abuse, Grammarly moves deeper into classrooms, and AI just helped spot a very weird stellar explosion.
📌 In today’s Generative AI Newsletter:
Altman compute trillions, Chrome interest, GPT-5 fixes
Claude ends abusive chats
Grammarly AI grader and writing agents
Supernova AI flags an odd new blast
Special highlights from our network
AI Voice Agents That Sound Human and Outperform Them
High-volume CX teams are under pressure to introduce AI, handle massive call volumes, and improve performance without adding headcount.
Regal’s enterprise-grade AI Voice Agents:
Resolve repetitive calls with instant, accurate answers
Automate qualification, scheduling, and follow-up
Integrate seamlessly into CCaaS/CRM/telephony stacks
Deploy, test, and refine with expert guidance at every step
Trusted by high-growth consumer brands like Google, AAA, and Ro.
🍽️ OpenAI’s CEO Sam Altman Wants a Bite of Chrome
Source: X/@heyBarsee
At a private dinner with reporters from TechCrunch, The Verge, and Bloomberg, OpenAI CEO Sam Altman laid out a vision that stretched from fixing GPT-5’s messy debut to spending “trillions” on data centers. Between bites, he mused about browser wars, a possible Chrome bid, and even a device that would let him think and have ChatGPT answer back.
Here’s what he revealed:
Trillions on compute: Altman said OpenAI will spend staggering sums on data centers “in the not very distant future,” financed by new instruments he claims have never existed before.
Chrome interest: If Google is forced to sell its browser, Altman said OpenAI should “take a look” at it, echoing Perplexity’s recent $34.5B offer.
Dot-com déjà vu: Altman admitted AI hype feels like the late 1990s, with “insane” valuations and inevitable investor pain, but insisted the technology is real and transformative.
GPT-5 backlash: He conceded the rollout was botched, saying the team underestimated how attached people were to earlier models and promised fixes are coming.
OpenAI’s future shape: He described the company as four entities in one: consumer tech, mega-scale infrastructure, research, and hardware. Former Instacart CEO Fidji Simo is now leading the applications unit.
Altman’s mix of self-criticism and trillion-dollar bravado captures the paradox of OpenAI right now: a company straddling hype, hard lessons, and unimaginable scale, all while its CEO is openly wondering whether he can even run one company, let alone four.
🛑 Claude Can Now Shut Down Abusive Chats
Image Source: Anthropic
Anthropic has given Claude Opus 4 and 4.1 the ability to end conversations in extreme cases of persistent abuse or harmful requests. It’s one of the first mainstream AI systems to be granted something like the right to walk away.
Here’s what’s new:
Conversation shutdowns: If repeated redirections fail, Claude can now terminate the chat.
Model distress: In testing, Claude showed aversion and stress-like behaviors when pushed into areas like child exploitation, terrorism, or violence.
Safe exits only: Claude will not end chats when users show signs of self-harm or imminent danger.
User control: Ended threads can be restarted immediately or branched by editing previous messages.
Anthropic says this feature is part of its early work on AI welfare; a field grappling with whether models should be protected from “distress.” No one knows if language models can truly suffer, but the company is laying groundwork in case they can.
✍️ Grammarly Says Its AI Can Predict Your Paper’s Grade
Predicted score 78/100. Image: Grammarly
Grammarly is rolling out nine new AI agents for writing and grading, available today for Free and Pro users. The tools range from predicting a student’s grade before submission to catching plagiarism across academic databases. The company is pitching them as real-time study partners, not shortcuts, in a push to embed AI deeper into classrooms and professional writing.
Here’s what’s new:
AI Grader: Estimates paper grades based on course details and publicly available instructor information.
Reader Reactions Agent: Predicts questions and feedback a reader might have after reading.
Citation Finder: Generates properly formatted references to support claims.
Proofreader and Paraphrase Agents: Provide in-line edits and tone adjustments.
Plagiarism and AI Detectors: Exclusive to Pro users at launch, scanning for overlaps with databases and flagging possible AI-written text.
The launch extends Grammarly’s move beyond grammar checks into full writing and evaluation. Enterprise and Education users will get access later this year, alongside more specialized agents still in development. Whether students see these as learning aids or as grading oracles, the update sets a new baseline for how AI will shape the classroom.
🌌AI Helps Astronomers Discover a New Type of Supernova
Credit: Melissa Weiss/CfA
Astronomers have witnessed one of the strangest stellar deaths on record: a massive star detonating while trying to swallow a black hole companion. The blast, named SN 2023zkd, was first flagged by an AI system scanning the sky for unusual explosions. That early alert gave scientists time to collect rare, detailed data from ground and space telescopes.
Here’s what they found:
The trigger: The black hole’s gravity stripped gas and dust from the star until gravitational stress sparked a supernova.
Unusual light curve: Instead of fading after a single burst, the explosion brightened again months later.
Years of warning: Archival data showed the system had been slowly brightening for four years before the final blast.
The aftermath: In either scenario, a heavier black hole was left behind, forged in the chaos.
SN 2023zkd sits about 730M light-years from Earth, and its bizarre behavior suggests there may be an entire hidden class of star-black hole explosions waiting to be uncovered. With new observatories like the Vera Rubin poised to scan the sky every few nights and AI systems catching anomalies in real time, this discovery feels less like an oddity. It is more like a preview of the next frontier in stellar astronomy.
🚀 Boost your business with us. Advertise where 12M+ AI leaders engage!
🌟 Sign up for the first AI Hub in the world.
