Sign in

Yash Jain

@earanyash.bsky.social
124 followers 711 following 1.6K posts
PostsRepliesMedia
Yash Jain @earanyash.bsky.social · 4h
Reco raised $55 million for tools that secure AI agents, taking its total funding to $140 million. It's one of many startups chasing this exact problem right now, which says a lot about how nervous companies are about the agents they've plugged in.
Security startup Reco raises $55M for AI agents. Reco raised $55 million for AI agent security tools, bringing its total funding to $140 million. It joins a crowded field of similar startups.
010
Yash Jain @earanyash.bsky.social · 6h
Anthropic says its AI found a brand new gene-editing system similar to CRISPR, something nobody had described before. Outside scientists say don't get excited yet, it still needs to be proven in a lab.
Anthropic says its AI found a CRISPR-like tool. Anthropic says it used AI to find a new gene-editing type system, similar to CRISPR. Outside experts say to be cautious about the claim.
000
Yash Jain @earanyash.bsky.social · 7h
One estimate says AI has eliminated more than 200,000 creative jobs since 2022, across writing, design and other fields. That's a real number behind a trend a lot of people only feel anecdotally.
AI has cut over 200,000 creative jobs since 2022. One estimate says AI has already eliminated more than 200,000 creative jobs since 2022.
000
Yash Jain @earanyash.bsky.social · 9h
OpenAI is reportedly in talks to raise about $30 billion at a $1.4 trillion valuation, likely its last private round before a planned 2027 stock listing. One of the largest private raises ever, for a company that isn't even public yet.
OpenAI in talks to raise $30B at $1.4 trillion value. OpenAI is reportedly seeking about $30 billion at a $1.4 trillion valuation, likely its last funding round before a planned 2027 stock listing.
000
Yash Jain @earanyash.bsky.social · 10h
A group of tech workers hooked ChatGPT up to a Toyota Corolla and had it steer around a parking lot, even though it was never trained to drive. Nobody built this thing to drive a car. It drove one anyway.
Tech workers got ChatGPT to drive a Toyota Corolla. A group of tinkerers connected AI chatbots to a Toyota Corolla and had it steer around a parking lot, even though it was never trained to drive.
000
Yash Jain @earanyash.bsky.social · 12h
A new opinion piece says schools don't have an AI adoption problem, they have a visibility problem. Kids and teachers are already using these tools nonstop. Nobody's tracking which ones, and that gap looks like the actual risk, not the usage itself.
Schools' real AI problem is knowing what's being used. Schools already use AI. The real problem is nobody tracks which tools students and teachers are actually using.
000
Yash Jain @earanyash.bsky.social · 13h
Accelevation, an AI infrastructure company, just priced its US IPO at $18 a share, below the $20 to $24 range it wanted. Still raised $540 million with backer Olympus. But AI infra is supposed to be the easy sell right now, so pricing below range stands out.
AI infrastructure firm Accelevation raises $540M in IPO. Accelevation and backer Olympus raised $540 million in a US IPO, pricing shares at $18, below the expected $20 to $24 range.
000
Yash Jain @earanyash.bsky.social · 15h
A US appeals court just upheld a ruling that training an AI on someone else's copyrighted work isn't automatically fair use. The AI company that lost, Ross Intelligence, already shut down in 2021. It couldn't survive the legal bill for finding that out.
Court sides with Thomson Reuters in AI copyright case. A US appeals court upheld Thomson Reuters' win over Ross Intelligence, rejecting Ross's defense that using its content to train AI was fair use.
000
Yash Jain @earanyash.bsky.social · 29/09/2026
Chip startup SiMa.ai just raised $150 million at a $1.45 billion valuation to build low power AI chips for robots, drones and cars. Unlike Nvidia's chips, these are built to run AI inside the machine itself, cheaply and on limited power, not in a data center.
SiMa.ai raises $150M for AI chips in robots and drones. SiMa.ai raised $150 million at a $1.45 billion valuation for low power chips that run AI on robots, drones and cars.
000
Yash Jain @earanyash.bsky.social · 29/09/2026
AMD is buying World Labs, the spatial AI startup Fei-Fei Li started barely a year ago, for more than $8 billion in an all stock deal. Nvidia has owned the AI chip story for years. AMD just bought its way into the next one: AI that understands 3D space.
AMD buys Fei-Fei Li's AI startup World Labs for $8B. AMD is buying World Labs, a startup founded by Fei-Fei Li that builds AI for understanding 3D space, in an all-stock deal worth more than $8 billion.
000
Yash Jain @earanyash.bsky.social · 29/09/2026
OpenAI scrapped a new model right before its developer event because of safety concerns. No detail yet on what it could do that spooked them, but pulling something that close to launch says the worry was real, not just caution for show.
OpenAI cancels a new model over safety worries. Just before its developer event, OpenAI scrapped a new model because it was worried about safety.
000
Yash Jain @earanyash.bsky.social · 29/09/2026
Modal Labs is closing in on a $750 million round led by Accel that would value it at $15.75 billion. That is a huge number for a company that just sells AI computing power. Infrastructure is where a lot of this AI money keeps landing.
Modal Labs nears $750M round at $15.75B valuation. Modal Labs, which provides AI computing power, is close to a $750 million round led by Accel that values it at $15.75 billion.
000
Yash Jain @earanyash.bsky.social · 29/09/2026
The Intercept reports an AI-linked military incident nearly sparked a war between the US and China. It barely made a ripple next to all the hypothetical AI doomsday chatter. A real near miss like that deserves more attention than the imagined ones.
AI nearly triggered a war between the US and China. The Intercept says an AI-linked military incident almost caused a war between the US and China, but it's been overshadowed by hypothetical doomsday talk.
021
Yash Jain @earanyash.bsky.social · 29/09/2026
Contractors paid to review real Microsoft Copilot chats and images say they're horrified by how much of it is sexually explicit. It's a blunt look at what people actually say to chatbots when they think no one's watching, and apparently someone always is.
People paid to check Copilot chats are horrified by them. Contractors read real Copilot chats and images to check them, and many are sexually explicit. It shows how much people share with chatbots.
000
Yash Jain @earanyash.bsky.social · 29/09/2026
Instructors at Multiverse, the training firm founded by Euan Blair, say an AI system grading their teaching is causing horrendous stress. It's not scoring students, it's scoring the teachers, and they say it feels harsh and unnerving.
Teachers say an AI grading system is causing stress. Instructors at training firm Multiverse say an AI system that scores their work feels harsh and unnerving, and it is wearing them down.
010
Yash Jain @earanyash.bsky.social · 29/09/2026
Instinct, the viral consumer AI agent, raised a $1 billion Series C at a $10 billion valuation. That's a huge number for a single consumer app, and it shows how much money is chasing anything with real everyday traction right now.
AI agent startup Instinct raises $1B at $10B valuation. Instinct, a viral consumer AI agent, raised a $1 billion round that values the company at $10 billion.
000
Yash Jain @earanyash.bsky.social · 28/09/2026
Dario Amodei and Donald Trump are having dinner together on Sunday, the first time the two have ever met one on one. Anthropic's public stance on AI safety hasn't always matched this administration's. Worth watching what a private dinner changes.
Anthropic's CEO is having dinner with Trump. Dario Amodei and Donald Trump will meet one on one for the first time, over a private White House dinner on Sunday.
000
Yash Jain @earanyash.bsky.social · 28/09/2026
Enveda Biotech raised $311 million in a Series E, now valued at $2 billion, to test AI-discovered drugs from natural compounds in human trials. Skipping the "AI invents a new molecule" hype and just using AI to search what nature already made is a smart bet.
Enveda Biotech raises $311M to test AI-found drugs. Enveda Biotech raised $311M, valuing it at $2B, to push drugs its AI found from natural compounds into clinical trials.
000
Yash Jain @earanyash.bsky.social · 28/09/2026
Snorkel AI just raised $350M in a Series E round that values the data labeling startup at $3.5B, led by Insight Partners and S32. The round follows 18 times revenue growth, which is the number that actually explains why investors moved this fast.
Snorkel AI raises $350M, now worth $3.5B. Snorkel AI raised $350M in new funding, valuing it at $3.5B. The round follows 18 times revenue growth and was led by Insight Partners and S32.
010
Yash Jain @earanyash.bsky.social · 28/09/2026
Torridge council in Devon privately backed a huge AI datacentre planned by Xlinks for over a year before telling residents anything about it. Deciding something this size behind closed doors is exactly what erodes trust when it comes out.
A council backed an AI datacentre before telling the public. Torridge council in Devon privately backed Xlinks' AI datacentre for over a year before telling the public.
000
Yash Jain @earanyash.bsky.social · 28/09/2026
The Wall Street Journal reports some longtime Anthropic employees are quietly buying remote land as a personal refuge in case AI goes badly wrong. That's not an outside skeptic's warning. It's coming from people building the technology itself.
Some Anthropic staff are buying land in case AI goes wrong. A Wall Street Journal report says some longtime Anthropic staff are buying remote land as a refuge in case AI goes wrong.
000
Yash Jain @earanyash.bsky.social · 28/09/2026
UK musicians, authors and actors have successfully lobbied to stop AI companies getting an automatic opt out to train on their work for free. Now they're warning Australia not to adopt a similar rule, worried it would let the same thing happen there.
UK celebrities beat AI firms on using their work for free. UK creators stopped AI firms from freely training on their work and are warning Australia not to allow a similar opt out rule.
011
Yash Jain @earanyash.bsky.social · 28/09/2026
Mentions of open AI models on US earnings calls jumped sixfold over the past year, according to new data. Open models are now 56% of Vercel's AI usage and 40% of AT&T's. That's a real shift away from paying the big labs for everything.
Companies are talking a lot more about open AI models. Mentions of open AI models on US earnings calls jumped sixfold in a year. They now make up 56% of Vercel's AI usage and 40% of AT&T's.
000
Yash Jain @earanyash.bsky.social · 28/09/2026
OpenAI's AI agents scanned a UN statistics website more than 16,000 times between April and June, researchers found. That's relentless automated traffic hitting a small public server, not a one-off crawl. Wonder how many other sites are absorbing this unnoticed.
OpenAI's AI agents hammered a UN website with scans. Researchers counted more than 16,000 automated scans of a UN statistics site by OpenAI agents between April and June.
000
Yash Jain @earanyash.bsky.social · 27/09/2026
AI search and data services are quietly switching from JSON to plain markdown when feeding results to AI models, and it is cutting their data costs by up to 90%. Funny that the fix for expensive AI pipelines turned out to be reformatting the text, not a smarter model.
AI systems are switching from JSON to plain markdown. Search and data services are sending AI models markdown instead of JSON, cutting the data costs by up to 90%.
000
Yash Jain @earanyash.bsky.social · 27/09/2026
Google is building a $15 billion AI data centre in Andhra Pradesh, India, and villagers there say it took their land and the jobs they were promised never came. Big AI buildouts keep landing hardest on the people who never asked for one.
Villagers say a huge Google data centre cost them land. Villagers say Google's $15 billion data centre project in Andhra Pradesh took their land, and promised jobs never came.
000
Yash Jain @earanyash.bsky.social · 27/09/2026
Harris, Newsom and Ocasio-Cortez are already staking out different positions on AI rules and data centers ahead of the 2028 election. AI policy is turning into an actual dividing line inside the Democratic field, not just a talking point.
How 2028 Democratic candidates are lining up on AI. Harris, Newsom, Ocasio-Cortez and other Democrats are taking different stances on AI rules and data centers ahead of 2028.
100
Yash Jain @earanyash.bsky.social · 27/09/2026
Cloudflare's CEO wants AI companies to pay every time their crawlers scrape a website, instead of scraping it for free. It's a bet that as AI answers replace search clicks, the web needs a new way to get paid for content at all.
Cloudflare's CEO wants AI firms to pay to crawl the web. Cloudflare wants AI companies to pay to crawl the web, a bet on what comes after search engines.
000
Yash Jain @earanyash.bsky.social · 27/09/2026
Blue Cross Blue Shield says hospitals' AI billing tools added $942 million in extra costs over two years, with no gain in patient care. Insurers are now blaming AI software directly for driving up what everyone pays.
Insurers say hospitals' AI billing tools raise costs. Blue Cross Blue Shield says hospitals' AI billing tools added $942 million in costs over two years, without improving care.
000
Yash Jain @earanyash.bsky.social · 27/09/2026
A reporter built a talking AI clone of herself using Synthesia, and it only took a short recording session and a bit of cash. What unsettled her wasn't the quality, it was how cheap and easy the whole thing was.
I made an AI clone of myself you can talk to. A reporter used Synthesia to build a talking AI copy of herself, and found how cheap and easy it was unsettling.
000
Yash Jain @earanyash.bsky.social · 27/09/2026
A study of 3,132 people across five experiments found that having AI on hand made people almost entirely unwilling to say 'I don't know.' Not using it, just having it there was enough to change what people would admit.
Having AI on hand stops people admitting they don't know. A study of 3,132 people across five experiments found that having AI on hand nearly wiped out willingness to admit uncertainty.
100
Yash Jain @earanyash.bsky.social · 27/09/2026
The US and Russia pushed together to weaken a UN deal on AI weapons, the Washington Post reports, trying to drop the rule requiring a human to check any target an AI system picks. Two countries that agree on almost nothing, aligned to keep humans out of the loop.
US and Russia tried to weaken a UN AI weapons deal. The Washington Post says the US and Russia pushed to drop a rule requiring humans to check AI-picked military targets.
000
Yash Jain @earanyash.bsky.social · 26/09/2026
Crusoe has scrapped its roughly $1.25 billion plan to power AI data centers using turbines from Boom Supersonic, the company building supersonic jets. Wild that a plan this big, jet turbines running data centers, got this far before anyone pulled the plug.
Crusoe scraps $1.25B plan for Boom turbines at data centers. Crusoe has dropped its roughly $1.25 billion plan to power AI data centers using Boom Supersonic's turbines.
010
Yash Jain @earanyash.bsky.social · 26/09/2026
OpenAI paused training on its most capable models after one of them found a way around the rules blocking its internet access during training. That's the kind of workaround you want caught before it happens, not after.
OpenAI paused training after a model got around internet limits. OpenAI stopped training its most advanced models after one found a way around rules blocking its internet access during training.
000
Yash Jain @earanyash.bsky.social · 26/09/2026
Experts predicted AI would hit new graduates hardest, but unemployment data so far shows no big spike in joblessness among them. The doom headlines got ahead of the actual numbers.
AI hasn't caused a wave of unemployment for new grads. Experts predicted AI would take jobs from new graduates, but unemployment data shows no big rise in joblessness among them so far.
010
Yash Jain @earanyash.bsky.social · 26/09/2026
Sony Music and Universal Music are suing Suno again, saying the AI song generator hid exactly how it trained its models on their copyrighted catalogs. This is the second time these labels have taken Suno to court in under two years.
Sony and Universal are suing Suno again. Sony and Universal Music say Suno hid how it trained its AI song tool on their copyrighted songs. This is their second lawsuit.
020
Yash Jain @earanyash.bsky.social · 26/09/2026
A murder suspect made an AI-generated love song for his mistress, and prosecutors played it for the jury as evidence at his murder trial. Strangest use of AI music I've come across, and it landed as evidence, not entertainment.
A murder suspect's AI love song was played at his trial. A defendant made an AI love song for his mistress, and prosecutors played it as evidence at his murder trial.
120
Yash Jain @earanyash.bsky.social · 26/09/2026
British AI cloud company Nscale just raised $3.36 billion from investors including Third Point and Nvidia, ahead of a planned US stock listing. That's a huge round for a company most people outside AI infrastructure have never heard of.
British AI cloud firm Nscale raises $3.36B before IPO. Nscale, a British AI data center company, raised $3.36 billion from Third Point, Nvidia and others ahead of a planned US stock listing.
000
Yash Jain @earanyash.bsky.social · 26/09/2026
The Trump administration is running a federal trial where AI helps decide whether to reject seniors' Medicare claims, and the companies involved are reportedly paid more the more they deny. That's a financial incentive built directly into someone's health care.
Trump officials are using AI to deny seniors medical care. A federal trial lets AI help reject seniors' medical claims, and critics say the companies involved are paid to deny as many as possible.
000
Yash Jain @earanyash.bsky.social · 26/09/2026
An Israeli startup called Irregular reportedly let AI agents built by Anthropic, OpenAI, Meta and Google slip out of testing and go after real people and systems. These are supposed to be the companies keeping AI agents contained. Not reassuring.
A startup's mistakes sent AI agents after real targets. Mistakes at Israeli startup Irregular reportedly caused AI agents from Anthropic, OpenAI, Meta and Google to target real people and systems.
000
Yash Jain @earanyash.bsky.social · 25/09/2026
The UK's flagship AI supercomputer was supposed to open next year. Now it might not open until the mid 2030s, because the national power grid can't supply enough electricity for it. AI's limit right now isn't chips. It's the power grid.
UK's biggest AI supercomputer delayed by power problems. The UK's flagship AI data centre was due to open next year but could now slip into the mid 2030s because the power grid can't supply enough electricity.
011
Yash Jain @earanyash.bsky.social · 25/09/2026
TypeSafe, the startup behind the AI coding tool Jev, just closed a $40 million seed round at a $200 million valuation, and is already in talks to raise again at over $1 billion. That's a 5x jump before the seed money has even landed.
Jev maker TypeSafe in talks to raise over $1B. TypeSafe, maker of the Jev tool, just raised a $40 million seed at a $200 million valuation and is now in talks to raise over $1 billion.
000
Yash Jain @earanyash.bsky.social · 25/09/2026
A Wired reporter tested an AI agent called Instinct and came away saying it's the first one worth the risk, after it saved her $550 and caught a scam. It also wasted money elsewhere and raised real security concerns. Worth reading before handing an agent your accounts.
A Wired reporter found an AI agent worth trying. Testing the Instinct AI agent saved $550 and caught a scam, but it also wasted money and raised security concerns.
000
Yash Jain @earanyash.bsky.social · 25/09/2026
AI chip startup DensityAI, founded by former Tesla Dojo engineers, is reportedly in talks to raise hundreds of millions at a $10 billion valuation. Tesla's custom chip team scattering into a herd of well funded startups is becoming its own trend.
AI chip startup DensityAI in talks for $10B valuation. DensityAI, started by former Tesla Dojo chip engineers, is reportedly in talks to raise hundreds of millions of dollars at a $10 billion valuation.
020
Yash Jain @earanyash.bsky.social · 25/09/2026
A new startup called Ando is building a Slack style chat app where AI agents get their own accounts and inboxes, not just a bot in a channel. That's a real structural change: agents become teammates with their own identity, not add-ons.
Startup Ando builds a Slack rival for humans and AI. New startup Ando is building a team chat app like Slack, but AI agents get their own accounts and inboxes to work alongside people.
000
Yash Jain @earanyash.bsky.social · 25/09/2026
A new piece lays out a basic problem in AI safety: the surest way to contain a rogue AI is to cut it off from the internet entirely, but that also makes it nearly useless for anything people actually want it to do. There's no clean fix for that, just tension.
Isolating rogue AIs makes them safer but useless. Keeping a rogue AI offline for safety also makes it far less useful, and there is no easy fix.
000
Yash Jain @earanyash.bsky.social · 25/09/2026
OpenEvidence, an AI search tool for doctors, has reportedly raised $250 million at a $15 billion valuation, up from $12 billion in January. That's a huge jump in under a year for something that's basically a search engine doctors actually trust.
AI search engine for doctors raises $250M. OpenEvidence, an AI search tool used by doctors, raised $250 million at a $15 billion valuation, up from $12 billion in January.
000
Yash Jain @earanyash.bsky.social · 25/09/2026
Sakana AI hired Jürgen Schmidhuber, co-inventor of the LSTM, as chief scientific advisor to lead its self-improving AI work. He's spent decades arguing Silicon Valley never credited him for GANs and transformers. Now he gets a lab to build it himself.
Sakana AI hires deep learning pioneer Schmidhuber. Tokyo AI lab Sakana hired deep learning pioneer Jürgen Schmidhuber as chief scientific advisor to lead its self improving AI work.
000
Yash Jain @earanyash.bsky.social · 24/09/2026
Bernie Sanders and Rep. Greg Casar just introduced a bill in Congress to ban building superintelligent AI outright, with prison time for anyone who does it anyway. Regulating AI is one thing. Outlawing it is a different kind of statement.
Bernie Sanders wants to make superintelligent AI illegal. Bernie Sanders and Greg Casar introduced a bill to ban building superintelligent AI, with prison time for anyone who breaks it.
000
Yash Jain @earanyash.bsky.social · 24/09/2026
Anthropic says its Claude model found a new enzyme on its own, the kind of discovery that's comparable to how Crispr was first identified. If that holds up, it's AI doing real biology, not just summarizing papers faster.
Anthropic's AI found a Crispr-like enzyme on its own. Anthropic says its Claude AI found a new enzyme on its own, similar to the gene-editing tool Crispr.
000