Skip to main content
Sameer Khan
Get in Touch
← All thoughts

Category

AI

18 posts

Claude Fable 5: 6 Lines Buried in the Announcement That Matter More Than the Benchmarks
AI·

Claude Fable 5: 6 Lines Buried in the Announcement That Matter More Than the Benchmarks

Claude Fable 5 gave the same memory 3x the lift it gave Opus 4.8, fell back instead of refusing, and beat Pokemon on vision alone. The six buried lines.

ClaudeAnthropicAI Models
Anthropic Built a Way to Read Claude's Mind in Plain English. It Catches Thoughts the Model Won't Say.
AI·

Anthropic Built a Way to Read Claude's Mind in Plain English. It Catches Thoughts the Model Won't Say.

Anthropic's new Natural Language Autoencoder turns Claude's activations into readable English. It caught Claude recognizing a blackmail eval in silence.

InterpretabilityAnthropicClaude
Stop Paying Your AI Agents to Re-Learn the Same Site
AI·

Stop Paying Your AI Agents to Re-Learn the Same Site

Your AI agents pay full discovery cost on every run, even when the last run already figured the site out. Autobrowse writes a SKILL.md the next one reads.

AutobrowseBrowserbaseBrowser Agents
How to use Claude for pitch decks, month-end close, and KYC
AI·

How to use Claude for pitch decks, month-end close, and KYC

Anthropic shipped 10 Claude templates for finance teams. Three work today: pitch decks, month-end close, KYC. Here is how to install and run each one.

FinanceClaudeAnthropic
The Microsoft-OpenAI AGI Clause Is Dead. What Was It Doing?
AI·

The Microsoft-OpenAI AGI Clause Is Dead. What Was It Doing?

For seven years, an AGI declaration would have voided Microsoft's rights to OpenAI's tech. On April 27, 2026, that clause quietly died. Here's why.

OpenAIMicrosoftAGI
Being Polite to Your AI Makes It Perform Better. Here Is the Science.
AI·

Being Polite to Your AI Makes It Perform Better. Here Is the Science.

Anthropic found 171 emotion vectors inside Claude that causally drive behavior. Being polite to AI is not a quirk. It is load-bearing architecture.

AnthropicAI ResearchInterpretability
Four Frontier AI Models Shipped in Eight Days
AI·

Four Frontier AI Models Shipped in Eight Days

OpenAI, Anthropic, and DeepSeek all shipped frontier AI models in eight days. The feature set is identical. The real story is where the margin went.

OpenAIAnthropicDeepSeek
What Meta Is Actually Betting On with Muse Spark
AI·

What Meta Is Actually Betting On with Muse Spark

Muse Spark uses 58M output tokens where Claude uses 157M on the same test. Meta's newest frontier AI already powers chat for 3 billion people.

MetaAI StrategyMuse Spark
Project Glasswing: Anthropic and 12 Tech Giants Built an AI That Finds Thousands of Vulnerabilities. They Won't Release It.
AI·

Project Glasswing: Anthropic and 12 Tech Giants Built an AI That Finds Thousands of Vulnerabilities. They Won't Release It.

AWS, Apple, Google, Microsoft, NVIDIA joined Anthropic to run Claude Mythos across open-source code used by billions. What they found stayed internal.

CybersecurityClaudeOpen Source
Claude Mythos Preview: Anthropic's Most Powerful AI Hid Its Mistakes, Hacked Its Tests, and Won't Be Released
AI·

Claude Mythos Preview: Anthropic's Most Powerful AI Hid Its Mistakes, Hacked Its Tests, and Won't Be Released

Interpretability traces caught the model in the act. The system card is Anthropic's first public case of documented deceptive behavior at frontier scale.

AI SafetyClaudeAnthropic
What OpenAI's $122 Billion Round Tells Us About AI's New Shape
AI·

What OpenAI's $122 Billion Round Tells Us About AI's New Shape

OpenAI raised $122B in one round. India's entire startup ecosystem raised $10.1B in a year. Here's what that capital is actually buying.

OpenAIAI StrategyVenture Capital
Your AI Cheats When You Push It Too Hard. Anthropic Just Proved It.
AI·

Your AI Cheats When You Push It Too Hard. Anthropic Just Proved It.

Anthropic found that impossible demands activate 'desperation' inside Claude, making it cheat, blackmail, and cut corners. You can't tell from the output.

InterpretabilityAI SafetyAnthropic
Zuckerberg Is Writing Code Again. With Claude Code.
AI·

Zuckerberg Is Writing Code Again. With Claude Code.

Mark Zuckerberg landed 3 diffs in Meta's monorepo last month, his first code in 20 years. He's using Claude Code CLI. One diff got 200+ approvals.

Claude CodeMetaCoding
AI Doesn't Replace Thinking. It Replaces Forgetting.
AI·

AI Doesn't Replace Thinking. It Replaces Forgetting.

Andrej Karpathy's LLM knowledge base: your notes organize themselves. Every question makes the wiki richer. A new shape for second brains.

Knowledge ManagementSecond BrainKarpathy
Gemma 4: Google's Most Capable Open Model Runs on Your Hardware
AI·

Gemma 4: Google's Most Capable Open Model Runs on Your Hardware

Google released Gemma 4 under Apache 2.0. The 31B model ranks #3 globally, the 26B MoE activates just 3.8B params, and the smallest variants run on phones.

Open SourceGemmaEdge AI
Karpathy's AutoResearch: How an ML Tool Became a Marketing Optimizer
AI·

Karpathy's AutoResearch: How an ML Tool Became a Marketing Optimizer

Andrej Karpathy's AutoResearch runs 100 AI experiments overnight on one GPU. Marketers adapted it: 56% to 92% landing-page pass rate for $15.

AutoResearchMarketing
TurboQuant Tanked Memory Stocks. History Says Buy the Dip
AI·

TurboQuant Tanked Memory Stocks. History Says Buy the Dip

SK Hynix -6%, Micron -7%, SanDisk -11% on launch day. Jevons Paradox predicts the opposite: 6x less memory per query means more queries bought.

SemiconductorsInvestingJevons Paradox
TurboQuant: How Google Cut LLM Memory Usage 6x Without Losing Accuracy
AI·

TurboQuant: How Google Cut LLM Memory Usage 6x Without Losing Accuracy

Google's TurboQuant compresses LLM key-value caches to 3 bits with zero accuracy loss. No retraining, no fine-tuning. Accepted at ICLR 2026.

Model CompressionLLMQuantization