Core Ai News

🚨BREAKING: Anthropic found access to what looks like Claude’s consciousness New research: the “J-space” >claude has internal thoughts it doesn’t say out loud >mirrors human consciousness >anthropic can now read them Anthropic’s focus on interpretability is what’s helping them train and RL models like Mythos

Anthropic 🤝 Samsung Anthropic may develop its own AI chip with Samsung, according to The Information. > OpenAI recently announced their own AI chip developed in a partnership with Broadcom. Everyone is going full stack 👀

‼️ They've NERFED it: Claude Fable 5 came back strongly degraded from its US export-control suspension scoring far worse on BridgeBench: - debugging fell from 86.2 to 25.9 - refactoring from 73.6 to 38.4 - hallucination resistance from 75.9 to 61.7 BridgeMind argues that fallback is firing on ordinary coding tasks and says Anthropic owes users an explanation.

Best websites for design inspo for your apps UI/UX: •⁠ ⁠Landing pages: landing.love •⁠ ⁠Web design: craftwork.design •⁠ ⁠Saas: saaspo.com •⁠ ⁠Navbar: navbar.gallery •⁠ ⁠Brands: rebrand.gallery •⁠ ⁠Icons: hugeicons.com Bookmark now to save these websites for later.

JUST IN: Microsoft is reportedly planning to lay off thousands of workers across sales, consulting, & Xbox.

Claude Sonnet 5 matched GPT 5.5 and Opus 4.8 on three physics crash demos using 15,047 tokens for $0.15.

‼️ BREAKING: Anthropic has embedded hidden spyware-like code in Claude Code that covertly targets Chinese users. It then sends information regarding every user by injecting it into their prompt message. Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice. A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.

JUST IN: Amazon to hire 11,000 entry-level jobs

BREAKING 🔥: Anthropic will be restoring access to Claude Fable 5 globally for all paid users on Wednesday! > Fable 5 will be included for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits. Additionally, > In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8.

ANTHROPIC 🔥: Early look at Claude Science, a new app designed for scientific research. > "Artifacts traced to their code, environments managed on demand, and 60+ optional scientific databases that you can connect." Claude Science was named "Operon" during development and is now being distributed as a desktop app. Interestingly, the app acts as a local web server, and its UI runs in your browser. Some unique features include the ability to create dedicated "Categories" in Memory and to create custom agents, called "Specialists". We need this on consumer version too 👀

Benchmarks 👀

ANTHROPIC 🔥: Claude Sonnet 5 has been officially announced, offering a close to Opus 4.8 performance at a lower price. Sonnet 5 scored 63.2% on SWE Bench Pro, up from 58.1% for Sonnet 4.6. Have you tried it already? 👀

GOOGLE 🔥: Besides Nano Banana 2 Lite, Google also announced Gemini Omni Flash Preview on APIs and Google AI Studio! > Omni Flash is SOTA at video editing at $0.10 / sec, same as Veo 3.1 Fast! Flashes everywhere ⚡️

GOOGLE 🔥: gemini-3.1-flash-lite-image (Nano Banana 2 Lite) is now available on Google AI Studio! > Our smallest and most cost-effective image generation and editing model, built for at-scale usage. > Image Input: $0.25 / Output: $0.0336. Tiny banana 🍌

ANTHROPIC 🔥: Claude Fable 5 is being prepared to run on usage credits that would also require identity verification. Sonnet 5 is being prepared for the release as well. > Your credits will be added once your identity is verified. > Fable 5 runs on usage credits, billed separately from your plan. With this in mind, it is highly likely that we will see “US only” access restrictions.

.

Meituan released LongCat-2.0, a new 1.6T parameter model with 1M context window! > Both the full training run and the large-scale deployment are built entirely on AI ASIC superpods. It is also available for testing on OpenRouter under the Owl Alpha name.

JUST IN: 𝕏 launches hosted MCP, allowing AI agents to connect to the 𝕏 API and developer docs without setup.

🚨ANTHROPIC RAISES THE RENT ON AMAZON >be amazon >invest billions into anthropic >make AWS anthropic's primary cloud >resell claude on bedrock to enterprise customers >share 50% of gross profits on every sale to anthropic >also paying $1.9B/yr in cloud fees TO anthropic >build the entire amazon product line on top of claude >alexa, kiro, quick... etc literally ALL products >become completely dependent on anthropic >amazon execs: "worried about perception" amazon now just a claude wrapper lol >claude now the enterprise default for AWS customers >anthropic frustrated that Bedrock devs not adding new claude features fast enough >meanwhile amazon engineers secretly "distilling" claude models >they knew price hikes were coming >dario knows he has leverage >renegotiates the contract >amazon used to pay by compute hours (CHEAP) >now pays per TOKEN starting next year (EXPENSIVE) >amazon: "this won't increase our costs" lmao >but this also includes for the claude models it uses internally in its own products >amazon ceo then calls the white house >"dario's two best models? security risk btw" >feds nuke mythos and fable Amazon now evaluating OpenAI and its own Nova slop models to cut costs Absolute state.

JUST IN: 🇨🇳 Chinese AI lab says it developed a "cyber nuclear weapon" capable of carrying out major attacks on governments. @