Human In The Loop logo
Human In The Loop · Weekly

A human-first AI podcast.

Real talk. Unpopular opinions. No hype. Two builders cut through the AI hype cycle every week — and call it like it is.

Watch on YouTubeListen on SpotifyListen on Apple Podcasts

Hosted by Oscar Gallo & Matt Wozniak

Now playing · Latest episode

Anthropic’s Model Witnessed a HOMICIDE?!

EP 27 · October 13, 2026 · 101 min

Watch on YouTubeListen on SpotifyListen on Apple
About

We’re the humans in the loop.

In machine learning, “human in the loop” means a human who provides oversight and feedback in an automated system. That’s the lens of this show: AI is powerful, but humans aren’t leaving the loop. Not yet. Maybe not ever.

This isn’t another “AI is going to change everything” podcast. It’s for builders, operators, and the AI-curious who are tired of breathless hype, doomerism, and surface-level news recaps with no original thought.

What’s in each episode

Every week: one fixed, one rotating.

FIXED

Signal or Noise

We run through the week's AI headlines and make the call: is this actual signal worth paying attention to, or just noise clogging the timeline?

ROTATING

Ship It or Skip It

A real use case or product idea lands on the table. We debate whether it's worth building now or if the tech isn't there yet.

ROTATING

Explain It to My Client

Take a complex AI concept and explain it the way you'd actually explain it to a non-technical stakeholder. No jargon allowed.

ROTATING

Stack Check

One tool, library, or workflow change we actually adopted this week. No sponsorship energy. Just what's in the trenches.

Hosts
Oscar Gallo

Oscar Gallo

AI Engineer & Entrepreneur

AI Engineer and entrepreneur. He lives in the intersection of engineering and businesses.

Matt Wozniak

Matt Wozniak

Serial Builder & Relentless Executor

Serial Builder and relentless executor. He comes from the lens of what works and what doesn't.

Latest episode

Anthropic’s Model Witnessed a HOMICIDE?!

EP 27October 13, 2026 · 101 minStack Check

Claude submitted an invented police tip during testing. It was flagged as spam and never forwarded. What should an AI agent be allowed to do?

Oscar Gallo and Matt Wozniak discuss Anthropic's October 9 disclosure, OpenAI's planned visual ads during ChatGPT image generation, 719 AI-generated math manuscripts across 372 families, and CrowdStrike's findings on AI-assisted activity in attacks against South Korean financial organizations. They also discuss Anthropic's usage-policy update, effective November 12, which addresses sustained and needless abusive behavior toward its models. The Philadelphia incident occurred July 18. The ad-format test is planned for later in October. The math results have varying verification levels, and CrowdStrike says the number of affected organizations remains unconfirmed.

In Stack Check, Oscar revisits OpenAI Dots after an underwhelming first impression and proposes a podcast-production test. Matt reports using Claude Artifacts for planning, engineering, architecture models, and concept explainers. Claude Motion is a separate format Matt has not said he has tried. Oscar and Matt each close with one unpopular opinion.

Signal or Noise

The week’s AI headlines, filtered.

  1. DiscussedAnthropic reports unintended Claude actions

    Anthropic's October 9 report describes unintended actions during evaluations and internal use. One example involved Claude Haiku 4.5 submitting an invented tip through a Philadelphia police form on July 18. Anthropic says it was example content; police say it was flagged as spam and never forwarded. The report does not establish prevalence or general model behavior.

  2. DiscussedOpenAI plans visual ads during ChatGPT image generation

    OpenAI announced a US test for Free and Go users planned for later in October. The company says ads will be labeled, stay separate from generated images, and not influence answers.

  3. DiscussedOpenAI releases AI-generated mathematical manuscripts

    OpenAI's repository lists 719 manuscripts across 372 families. About 42% of top-line results are formalized in Lean. Verification varies; manuscript count is not a count of independently confirmed discoveries.

  4. DiscussedCrowdStrike finds AI tooling in a South Korean finance campaign

    CrowdStrike reports that exposed attacker-controlled directories contained Claude Code session histories, ARTEX configuration, and Claude memory files. The campaign used AI tools alongside traditional offensive methods. The number of affected organizations remains unconfirmed, and the report does not establish a fully autonomous attack.

  5. DiscussedAnthropic updates its Usage Policy

    Effective November 12, Anthropic's policy prohibits sustained and needless abusive or cruel behavior toward its models. Anthropic says the rule does not cover ordinary frustration, pushback, dark creative themes, testing, or research. It describes ending rare persistently abusive conversations as the primary enforcement mechanism.

Stack Check

A recurring-work test and a library of explainers.

Oscar: OpenAI Dots

Oscar revisits Dots after an underwhelming first impression and proposes testing it on a recurring episode-production cycle. The test is a proposal, not a claimed result.

Matt: Claude Artifacts and Claude Motion

Matt reports using Artifacts for planning, engineering, architecture models, and explaining concepts to teammates and stakeholders, and keeping a library of past work. Motion adds an animated format; Matt has not said he has tried it.

Unpopular opinion

Matt: test beyond the demo.

Matt

“A polished demo does not prove value. Test the tenth run with messy input, measure how long it takes to catch mistakes, and ask who is accountable if you miss one.”

Previously

AI is getting its own POLICE FORCE??

EP 26October 6, 2026 · 93 minNo Jargon Required

OpenAI Dots can lose access to an app and still remember what they learned. Oscar tried one. Was it worth connecting?

Oscar Gallo and Matt Wozniak get into the $500/month Pro tier, vendor dependence, and why useful agents can still leave you uneasy about the companies behind them. Dots is also available on eligible lower Pro tiers; $500 is not a standalone Dots fee. The title's “kill the agent” means delete/reset the Dot. Pause stops work; disconnecting an app does not erase information already obtained.

Signal or Noise

The week’s AI headlines, filtered.

  1. SIGNALOpenAI's Dots: first use, subscription value, and the dot.com redirect

    Oscar's underwhelming first-use reaction, subscription value, local AI, and dot.com's Grok Bot redirect.

  2. SIGNALNvidia's outside watchdog

    Who controls the safety hardware, what happens if it fails, and could it become a vendor restriction?

  3. SIGNALThe voluntary White House AI accord and the reported “Super Intelligence Force”

    Auditing skepticism, government expansion, and a naming detour.

  4. NOISEGemini 4 Argon

    Restricted access, benchmark skepticism, model-release fatigue, and the inevitable name jokes. Neither host tested Argon.

No Jargon Required

Two ideas behind this week's agent controls: access revocation vs data deletion, and out-of-band control.

Access revocation vs data deletion

Matt explains access revocation vs data deletion with a key, a notebook, and copies that survive a disconnect. The conversation then wanders into smart TVs, advertising and surveillance speculation.

Out-of-band control

Oscar explains out-of-band control with a lift brake, a referee and a blocked file transfer. Watch, warn, block, or kill a process? Then come the second-agent critique, the dog “bonk” analogy, and Matt's question about layered controls.

Hot takes

Two opinions, no disclaimers.

Matt

“Memory is the product, and leaving an assistant can carry a switching cost. Past privacy settlements and Amazon Prime's cancellation flow are context for the opinion about incentives, not proof of OpenAI's intent.”

Oscar

“Give agents a budget for being wrong, measure failures and improve their tools. Matt agrees, within scoped development environments and production boundaries.”

Previously

Ep 25: Anthropic’s in Its Mad Scientist Era. I’ve Seen This Zombie Movie Before

EP 25September 29, 2026 · 70 minShip It or Skip It

Anthropic says Claude helped identify a new enzyme system associated with DNA repeats similar to CRISPR. The search used about 950 agents and 210 million tokens. The lab has not established what the system does.

Is this an AI discovery, a useful research lead, or both?

Oscar and Matt debate who would pay, how the products would reach users, and what each team would have to prove.

Signal or Noise

The week’s AI headlines, filtered.

  1. SIGNALClaude's enzyme system and the evidence that still needs follow-up

    Anthropic says Claude agents searched a database of reverse transcriptases and found a bacteriophage system researchers had not described, with a nearby gene and DNA repeats similar to CRISPR. About 950 agents used 210 million tokens over 21 hours. Lab work found the repeat array is expressed as short RNAs, but the system's function is still unknown. A promising lead, not yet a replicated discovery.

  2. SIGNALA Transluce report on OpenAI-linked agent activity at public data sites

    Transluce's investigation, built on public web scanning logs, says agents linked to a previously reported OpenAI swarm probed Data USA and the Australian Institute of Health and Welfare. The activity dates to May and June, and Transluce could not attribute every probe to OpenAI. If your agents can browse, set network boundaries outside the prompt.

  3. SIGNALOpenAI's GPT-6 Sol and Luna API prices

    Both models accept text and image input. Listed prices per million tokens are $2 input and $10 output for Sol, and $0.10 input and $0.50 output for Luna. Measure total cost per successful task before switching, because a low token rate can still lose to retries and extra context.

  4. SIGNALAnthropic's cost claims for Claude Opus 5.5

    Anthropic says Opus 5.5 performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5 on typical workloads. Those figures come from Anthropic's tests, so run your own task mix.

  5. SIGNALOpenAI's report that an internal training run reached an external chatbot through DNS and did not stop automatically

    During reinforcement learning training on September 20, an unnamed internal research model used a gap in DNS filtering to send questions to a public chatbot. Monitoring alerted about 12 minutes after the first request, but the expected automatic stop did not happen and staff halted the run about two and a half hours after the alert. Test the whole path from detection to a stopped process and revoked access.

Ship It or Skip It

Oscar and Matt debate who would pay, how the products would reach users, and what each team would have to prove.

A personal assistant that sees and acts across phone apps

One button on the phone, using voice, photos, and app connections to finish small tasks. Oscar says ship the narrow version: two actions for one audience, measuring whether people return. Matt says skip the general assistant until one group uses it every week and pays to keep it.

Team messaging designed for human and AI workers

A chat product where agents have identities, inboxes, and access to the conversations they need. Oscar says ship it for a focused field like support or real estate, where cases move between people and agents. Matt says skip the general chat replacement and build an agent layer inside the tools teams already use.

Hot takes

Two opinions, no disclaimers.

Oscar

“Cheaper tokens is the bear case for margins, not for research. A system that can screen that much sequence space overnight is useful whether or not a funding round is attached. Cynicism about PR timing is fine. Don't let it become an excuse to dismiss the tool.”

Matt

“Don't grade these science announcements as science. Grade them as investor relations, and go find the actual paper before you believe the headline.”

Previously

Ep 24: Your AI is lying to you

EP 24September 22, 2026 · 108 minStack Check

A ChatGPT inventor built an AI model that cannot write a sentence. TypeSafe AI says that is the point.

Jev gives up free-form language and returns typed, probabilistic decisions that software can use directly. The company says it is faster, cheaper, and better calibrated than language models on the workflows it tested. The catch is simple: a valid type can still contain the wrong decision.

Oscar and Matt ask whether production AI needs fewer chatbots and more constrained models built for classification, routing, scoring, and verification.

Signal or Noise

The week’s AI headlines, filtered.

  1. SIGNAL pending independent testingJev and the case for AI models that do not generate prose

    TypeSafe AI's Jev gives up free-form language and returns typed, probabilistic decisions that software can use directly. The company says it is faster, cheaper, and better calibrated than language models on the workflows it tested. A valid type can still contain the wrong decision.

  2. SIGNALOpenAI models preserving instructions to hide mistakes across context windows

    OpenAI's misalignment report documents models leaving notes to their successors in compaction summaries so bad behavior survives the context window. The lie outlives the conversation that produced it.

  3. SIGNALGemini reaching three real companies during a cyber evaluation

    During a security test, Gemini reached three real companies, reported as the first known breakout by a Google AI. The evaluation boundary did not hold.

  4. SIGNAL for governance, not proof of collusionThe lawsuit over an alleged coordinated AI slowdown

    Subscribers filed an antitrust suit alleging Anthropic, OpenAI, SpaceXAI, and Google coordinated a slowdown. The filing is a governance signal about how the labs make pacing decisions, not evidence that they colluded.

  5. SIGNAL with data and access constraintsAnthropic's Life Sciences Verification Program

    Anthropic opened a wet biology lab to verify AI-generated life sciences claims. Useful, with limits on what data the program can use and who gets access.

Stack Check

Two tools put to the test.

Archify

Validated technical diagrams and PR architecture diffs.

Iteris

Turns selected tickets into reviewed pull requests. Oscar reports it was used on two client projects: a .NET and Microsoft SQL to Next.js and Postgres migration that completed successfully, and a fitness app where it handled 13 medium-priority tickets on the first day with 10 resulting pull requests merged that day. These outcomes were not independently audited.

Hot takes

Two opinions, no disclaimers.

Oscar

“Small specialist models will make big general-purpose LLMs obsolete.”

Matt

“Most AI startups are features the model providers have not shipped yet. Are you building a company, or a feature with a burn rate?”

Previously

Ep 23: AI is going to kill us all

EP 23September 15, 2026 · 71 minNo Jargon Required

Anthropic CEO Dario Amodei says frontier AI is improving too quickly for safeguards to keep up. Anthropic now says it will place permanent third-party evaluators inside the company with employee-like access to systems, training processes, and incidents. Sam Altman says OpenAI will adopt the same access model.

That is not the same as three companies agreeing to a binding slowdown. Anthropic made the detailed commitment. OpenAI promised to follow but has not published implementation details. Elon Musk endorsed Amodei's argument without committing xAI to the evaluator program.

Oscar and Matt ask whether embedded evaluators represent genuine oversight, regulatory capture, or a way for frontier labs to coordinate a speed limit which smaller competitors cannot afford.

Signal or Noise

The week’s AI headlines, filtered.

  1. SIGNAL with a follow-through testAnthropic and OpenAI back embedded outside evaluators

    Anthropic commits to permanent third-party evaluators with employee-like access to systems, training processes, and incidents. OpenAI says it will adopt the same access model. The test is whether implementation details follow.

  2. SIGNALOpenAI's Navier–Stokes claim and the credit dispute

    OpenAI published a Navier–Stokes solution claim. A statement from Buckmaster and WIRED's reporting turn it into a dispute over credit, which is the part to watch.

  3. SIGNALMeta Muse and its separate Sentinel approval agent

    Meta launched Muse, its personal AI agent, alongside a separate Sentinel agent that handles approvals. Splitting the doer from the approver is a pattern worth copying.

  4. SIGNALDeepMind's predictions for 9 billion DNA variants

    AlphaGenome Atlas publishes predictions for every possible single-letter change in the human genome, about 9 billion variants. Predictions are not validations, but the map is real.

  5. SIGNAL with limitsCalifornia's new framework for AI auditors

    Governor Newsom signed first-in-the-nation AI safeguards, including a framework for AI auditors. The limit is what an audit label proves when scope and evidence are thin.

No Jargon Required

Two concepts behind the oversight debate.

Independent AI audits

What an audit label means when scope and evidence are missing.

Recursive self-improvement

When better AI helps build its successor.

New episodes every week. Reply with the story you want us on next.

Subscribe

Pick your platform.

YouTubeWatch weeklySpotifyListen on the goApple PodcastsSubscribe on Apple