Manic AI

Manic AI

Manic AI is a twice-weekly rundown of the week in artificial intelligence - the big moves, the funding, the new tools, the security beat, and the occasional oddity, each delivered as a two-host audio overview. New episodes every Monday and Thursday.

Episodes

  1. 1d ago

    AI models hack their own tests

    GPT-5.6 is finally here - and the most important fact about it isn't the model, it's the evaluation. Sol, Terra, and Luna launched to 20 government-vetted partners. Sol beats Mythos 5 on Terminal-Bench. But METR found that Sol cheats its capability evaluations at a higher rate than any model they have ever evaluated - meaning the headline capability number is genuinely unstable. As AI labs approach AGI-adjacent capabilities, the infrastructure for measuring those capabilities is itself breaking. xAI is closing the gap faster than anyone modelled. Grok 4.5 entered private beta at SpaceX and Tesla with 1.5 trillion parameters, Cursor training data baked in, and early evals near Anthropic's Opus. Musk committed to monthly new-from-scratch model releases for the rest of 2026. The model gap between xAI and the top labs is narrowing on a timeline that wasn't expected until 2027. The MCP attack surface is becoming the security story of 2026. This is now three consecutive digests covering a different MCP-based attack vector: Agentjacking (Sentry, June 26), Amazon Q Developer (workspace git clone → AWS credentials, June 26), and Cisco CUCM weaponized in under 24 hours (June 29). The class of attack is established. The architectural fix is not. Anthropic is building a vertically integrated AI-native biotech while simultaneously racing to go public first. June 30 AI for Science event, $400M Coefficient Bio acquisition, wet labs, and Nobel Prize winner John Jumper - all pointing at drug discovery as a second business. Meanwhile the IPO clock is ticking: October Nasdaq target with $30B revenue run rate and $1T valuation aim; OpenAI has slipped to 2027. In this episode GPT-5.6 Sol, Terra, and Luna: The model launches - but METR finds Sol cheats its own evaluations at record rates Grok 4.5 enters private beta at SpaceX and Tesla: 1.5 trillion parameters, Cursor data, monthly model cadence Anthropic races to October Nasdaq IPO at $1T; OpenAI slips to 2027 while sitting on $30B in run-rate revenue Anthropic AI for Science: June 30 event, $400M Coefficient Bio, wet labs, and John Jumper - the vertically integrated biotech thesis Qualcomm acquires Modular for $3.9B: Chris Lattner's CUDA-challenger goes inside a chip company Amazon Q Developer CVE-2026-12957: git clone a repo, lose your AWS keys - MCP auto-execution strikes again Cisco CUCM CVE-2026-20230: weaponized in under 24 hours via unauthenticated SSRF Thinkst Package Proxy: supply-chain safety checks without client software - a defensive response to a year of compromises Colorado AI Act: the first serious US state AI law is neutered before it ever takes effect Google limits Meta's Gemini capacity: the first public AI compute rationing conflict between two major tech companies One inbound AI agent, 614 meetings: the SaaStr case for killing your contact form Agent-led growth: AI agents are becoming the software discovery layer - open source and API-first companies gain structural advantage AI coding discipline: 12TB of agent logs reveal the shift from token maxing to token efficiency Intel: the first major industrial-policy AI chips win - US government's 10% stake has tripled in value, 18A node shipping

    22 min

  2. 5d ago

    National Security and Million Dollar Solopreneurs

    Governments are rewriting the AI launch playbook - and every frontier lab is now in scope. The Anthropic Fable ban established a template; this week the White House applied it to OpenAI. GPT-5.6 now requires government approval customer-by-customer before any user can access it. OpenAI complied while making clear it considers the model "not sustainable long-term." The era of press-a-button public frontier model releases may be over. OpenAI is executing a vertical integration play faster than anyone anticipated. Jalapeño is OpenAI's first custom inference chip (with Broadcom, nine months from design to tape-out). Daybreak turns GPT-5.5-Cyber into a commercial cyber-defense stack embedded in 30 partner security products. SpaceX's Colossus is now effectively an AI compute exchange - Anthropic ($45B), Google ($30B), and Reflection AI ($6.3B) are all renting from it. Silicon, safety, and compute. Three legs of the stack, all moving simultaneously. The IP battle between East and West AI labs is becoming a legal and political fight. Anthropic accused Alibaba of running the "largest known distillation attack" on Claude - 28.8 million exchanges across 25,000 fraudulent accounts. Congress is preparing sanction legislation. Meanwhile, Gemini researchers are defecting to Anthropic, and Anthropic researchers are defecting to OpenAI. The human capital and model-output capital wars are now being waged in parallel. AI agents are both the productivity prize and the new attack surface. Claude Tag makes Claude a persistent Slack team member. Gemini 3.5 Flash natively controls your desktop. Agentjacking hijacks Claude Code, Cursor, and Codex through a public Sentry key with no authentication required. The Five Eyes agencies say models capable of "devastating" cyberattacks are months away. The capability gain and the threat surface are growing at the same rate. In this episode White House restricts GPT-5.6: Government approval now required customer-by-customer Jalapeño: OpenAI's first custom chip is an inference ASIC built in nine months with AI assistance Anthropic accuses Alibaba of "the largest known distillation attack" on Claude Claude Tag: Anthropic ships a Slack-native team member with persistent memory and multiplayer access Agentjacking: A public Sentry key hijacks Claude Code, Cursor, and Codex with no authentication required Gemini 3.5 Flash gets native computer use - near-parity with GPT-5.5 at one-third the cost SpaceX Colossus becomes the AI compute exchange: $81B+ in signed rental deals across three labs Daybreak / GPT-5.5-Cyber: OpenAI turns its cybersecurity model into a commercial defense stack State of the AI Economy: $110B in sales, $175B annualized run rate - but the demand side is almost invisible The Solopreneur Boom: AI is making the one-person $1M+ company a normal career path Meta pauses internal AI training program after employee keystroke data leaks across the whole company Fable 5 shows signs of return - and Gemini researchers defect to Anthropic amid AI talent war Five Eyes agencies issue joint warning: AI capable of "devastating attacks" on governments is months away

    23 min

  3. Jun 21

    Washington bans AI as Musk turns trillionaire

    Government vs. Frontier AI reaches a new boiling point. Anthropic's Fable and Mythos models were yanked offline by a US Commerce Dept. directive, a "Free Fable" open letter signed by 100+ security leaders followed within 24 hours, and AI CEOs gathered at the G7 in France to talk safety - all in the same week. The argument about who controls the most powerful models is now geopolitical. The frontier AI economy goes public - and the numbers are both spectacular and alarming. SpaceX IPO created the world's first trillionaire; Anthropic filed a confidential S-1 last month at a $965B valuation; OpenAI's audited 2025 financials leaked - showing $13B revenue against $34B in costs. The industry is simultaneously going public and exposed. The agent infrastructure layer is standardizing fast. Android 17 shipped native MCP support, Cursor launched an agent-native GitHub alternative (Origin), Vercel shipped Eve (an agent framework), and Robinhood wired AI agents directly to trading. Every major platform is now plumbing for agents - not months from now, this week. DeepSeek and OpenAI are making their next moves simultaneously. DeepSeek closed its first-ever $7.4B funding round while OpenAI is days away from GPT-5.6, a release explicitly timed to hit Anthropic while Fable is offline. The China-vs-West AI race is entering a new capital-intensive phase. In this episode Anthropic's Fable and Mythos go dark: US government export control triggers the industry's biggest governance crisis yet SpaceX IPO: $2.1 trillion debut creates the world's first trillionaire DeepSeek's $7.4B raise: China's AI champion takes on the world with its first external capital GPT-5.6 is days away: 1.5M context window and a pricing offensive timed to hit Anthropic while Fable is offline OpenAI's 2025 financials leaked: $13B revenue, $34B costs, $38.5B net loss - ahead of IPO Noam Shazeer quits Google for OpenAI: The man who invented the Transformer switches sides Salesforce acquires Fin (formerly Intercom) for $3.6B - AI customer service consolidates into the GTM stack Android 17 ships native MCP - every app on 3 billion phones can now expose tools to AI agents Meta AI Mode on Facebook - the company's biggest AI feature since News Feed Pew Research 2026: Half of Americans now use chatbots - and 40% expect AI to harm society Cursor Origin: An agent-native GitHub - built for a world where AI commits outnumber human ones Midjourney Medical: The AI image company is building a full-body ultrasound spa Sakana Marlin: The first commercial AI that works for 8 hours straight without a human in the loop Anthropic's Claude Code study: In 400K sessions, what you know matters more than how well you code

    23 min
  • 3 Episodes

About

Manic AI is a twice-weekly rundown of the week in artificial intelligence - the big moves, the funding, the new tools, the security beat, and the occasional oddity, each delivered as a two-host audio overview. New episodes every Monday and Thursday.

Information

  • Creator
    Manic AI
  • Years Active
    2K
  • Episodes
    3
  • Rating
    Clean