OpenAI just launched GPT-5.4-Cyber, its first model built specifically for cybersecurity defense.
It's a fine-tuned version of GPT-5.4 with fewer restrictions for legitimate security work, designed to give defenders access to frontier AI capabilities without the usual guardrails getting in the way.
It's a fine-tuned version of GPT-5.4 with fewer restrictions for legitimate security work, designed to give defenders access to frontier AI capabilities without the usual guardrails getting in the way.
👾47
So Anthropic leaked a full-stack app builder inside Claude.
> Lovable just raised money.
> Replit just shipped new features.
> Bolt just found product-market fit.
Imagine building a whole startup just for an AI lab to ship your product as a Tuesday update.
@CoreAti
> Lovable just raised money.
> Replit just shipped new features.
> Bolt just found product-market fit.
Imagine building a whole startup just for an AI lab to ship your product as a Tuesday update.
@CoreAti
👾46
Grok now supports uploading up to 100 files at once.
A super useful feature for batch analysis.
@CoreAti
A super useful feature for batch analysis.
@CoreAti
👾49
Core Ai News
So Anthropic leaked a full-stack app builder inside Claude. > Lovable just raised money. > Replit just shipped new features. > Bolt just found product-market fit. Imagine building a whole startup just for an AI lab to ship your product as a Tuesday update.…
Anthropic shipped 69 updates over 12 weeks.
👾48
👾49
📰 Everything else in AI today.
AWS rolled out Amazon Bio Discovery, a new drug-design platform with biological foundation models and a built-in lab network for synthesis and testing.
UK AI safety evaluators said Claude Mythos Preview is the first AI to complete their 32-step corporate hack simulation, showing big jumps in cyber-attacks over Opus 4.6.
Baidu released ERNIE-Image, an 8B open-weight text-to-image model that nears top rivals on benchmarks despite its small size.
OpenAI's Greg Brockman framed AI as the shift to a "compute-powered economy," claiming ~1B weekly ChatGPT and Codex users on the company’s 10th anniversary.
AWS rolled out Amazon Bio Discovery, a new drug-design platform with biological foundation models and a built-in lab network for synthesis and testing.
UK AI safety evaluators said Claude Mythos Preview is the first AI to complete their 32-step corporate hack simulation, showing big jumps in cyber-attacks over Opus 4.6.
Baidu released ERNIE-Image, an 8B open-weight text-to-image model that nears top rivals on benchmarks despite its small size.
OpenAI's Greg Brockman framed AI as the shift to a "compute-powered economy," claiming ~1B weekly ChatGPT and Codex users on the company’s 10th anniversary.
👾47
Trending AI Tools
🤖 Claude Code - Anthropic's revamped desktop app for parallel agents
📱 Skills in Chrome - Reusable prompt shortcuts for Gemini in Chrome
⛵️ Amazon Bio Discovery - Agentic app for antibody drug design, lab testing
Please open Telegram to view this post
VIEW IN TELEGRAM
👾48❤2
Google released Gemini 3.1 Flash TTS with support for 70 different languages!
> Available via a new audio playground in AI Studio and in the Gemini API!
@CoreAti
> Available via a new audio playground in AI Studio and in the Gemini API!
@CoreAti
👾47
Donald Trump believes we need to create a "kill switch" for all AI due to it posing a possible existential threat to humanity.
👾48
At risk of bankruptcy, shoe company Allbirds decided to sell everything and funnel all their money into AI computing in a hail mary pivot to save the company.
Their stock is up over 600% after making the announcement today.
Their stock is up over 600% after making the announcement today.
👾46😁1
⚡️ Google DeepMind launched Gemini Robotics-ER 1.6, a model enabling robots to understand space, tools, and task progress with fewer human checks.
Since vision alone is insufficient, the model uses stepwise reasoning, pointing, counting, multi-view success detection, and spatial constraints to convert camera feeds into decisions.Source.
Its standout feature is instrument reading: agentic vision zooms, marks indicators, runs code for proportions, and reads gauges at 93% accuracy. Demos with Boston Dynamics Spot show the system checking doors, reading dials, correcting distortion, and refusing risky actions like handling liquids or heavy items.
DeepMind also offers an API to integrate this reasoning into industrial tasks such as pallet counting, inspection, and puddle detection.
👾47❤1
🚨GPT-5.4 Pro just made a contribution to the field of mathematics.
to those interested:
> gpt-5.4 Pro solved a 60 year-old mathematical problem in just 80 minutes. it then took another 30 minutes to convert the full solution to a clean LaTeX paper.
to those interested:
> gpt-5.4 Pro solved a 60 year-old mathematical problem in just 80 minutes. it then took another 30 minutes to convert the full solution to a clean LaTeX paper.
this was an open question posed in 1966 on the asymptotic behavior of large primitive sets, where no element divides another. no one had cracked it in sixty years because every approach since Erdős in 1935 had shifted to probabilistic methods.
GPT 5.4 stayed purely in analytic number theory, working directly with the von Mangoldt function to encode the unique factorization identity.
we’re now in the era where AI solves century old problems 🥶
Grok needs to step up his game. they have the raw power to solve the universe’s mysteries with colossus.
👾46❤1
Anthropic
🚨Claude launched Opus 4.7!!
Claude Mythos and Opus 4.7 Benchmark
Mythos is ~10–15% better overall in real-world use.
Comparison:
Mythos is ~10–15% better overall in real-world use.
Comparison:
1👾48