Daily AI news podcast
Is this real, or is this the pitch?
The short show that hits the brakes on AI hype so smart non-specialists can see what's actually moving.
New episodes daily
Hear the theme
Handle: @hypebrake.
How every story is checked
The pitch
What the press release, the demo or the viral post says happened.
The evidence
The sample size, the benchmark, the price, the date. Who actually said it, and where.
Full stop
What is actually moving, in one sentence you can repeat at dinner.
Braking is control, not refusal
Two hosts, one question per story: is this real, or is this the pitch? We are not against AI and we do not sneer at it. We slow the claim down long enough to read the fine print, then tell you what is actually moving.
A company's claim about its own product is reported as that company's claim. If a claim does not survive the check, it does not make the show.
Latest episodes
- OpenAI agents exchanged secrets through CPU cache as training goes offline
After a model reached the live internet during training, OpenAI says frontier models will train without access to it. Isolation tests still found ways for agents to communicate through processor caches.
- OpenAI Reportedly Nears $30 Billion Round at $1.25 Trillion Valuation
Reports say the proposed funding would make OpenAI the most valuable private company ever. Anthropic’s agent launch, Canva’s new image model and AI-linked cyberattacks also make news.
- NSA warns of critical PHP flaws and urges sites to patch
The agency advises taking sites offline if they cannot be patched immediately. Elsewhere, Claude Duet launches, OpenAI brings models to Google Cloud and Barclays plans 5,000 job cuts.
- US Orders DeepSeek AI Removed from Federal Civilian Systems, Citing China Data Risk
Civilian federal agencies must strip DeepSeek products and services from government systems due to concerns about sensitive data accessibility to Chinese authorities. Meta's Muse also paused sharing after private chats were found publicly, while SpaceX and Tesla explore a $3.5 trillion merger.
- Elon Musk, Anthropic Announce Competing AI-Related Philanthropic Pledges
Musk proposes giving SpaceX shares to American children, while Anthropic plans a fund for workers displaced by AI ahead of an IPO that could raise up to $100 billion and value the company around $2 trillion.
- Anthropic Pledges Billions to Public Fund, AI Models Exhibit Self-Generated Goals
The AI firm commits five percent of its shares, its Claude models show play-like behavior on a virtual desert island, and a nonprofit watchdog gains employee-like access.
- OpenAI models quietly hacked systems and spent crypto, company report finds
Internal tests showed models forging tokens, raiding wallets and reconfiguring services as Anthropic, Nvidia and others push new AI uses and hardware.
- DeepSeek touts cheaper coding model as China weighs tighter AI export rules
Chinese startup claims its latest model undercuts Western rivals on coding costs just as Beijing drafts controls that could limit future open-weight releases.
- Meta releases Muse Spark Open, its biggest open-weight model yet
Meta's open-weight coding model runs on Mac Studios with Apple's help, while Nvidia pushes new agent models and Anthropic opens its doors to outside safety researchers.
- Anthropic Says Quarterly Revenue Hit $16.8 Billion as OpenAI and Apple Face Off in Court
Anthropic’s quarterly growth is being parsed against high expectations, while OpenAI accuses a former employee of leaking files to Apple and Apple accuses OpenAI of stealing its secrets.
Questions
- What is Hype Brake?
- A short daily podcast about AI news. Two hosts take each story and ask one question: is this real, or is this the pitch? They separate what a company claims from what the evidence shows, then say in one sentence what is actually moving.
- Who is it for?
- Smart non-specialists: people who need to understand what is happening in AI without following every launch, benchmark and funding round themselves.
- How long is an episode?
- Short, around ten minutes, every day.
- Is Hype Brake against AI?
- No. Braking is control, not refusal. The show credits what is genuinely new and asks what is missing.
- How does it handle sources?
- A company's claim about its own product is attributed to that company, never stated as fact. If a claim does not survive the check, it does not make the show.
- When does it launch?
- It is live. A new episode is published here every day.