CheckMates Go

The AI news segment contains a reasonable volume of concrete, current information (Evil Tokens stats, Anthropic/NSA red team story, AI coding agent detection methodology), but the first half of the episode is dense with generic cybersecurity pain-point narration that adds little beyond what any practitioner already knows. The newsletter-reading format limits synthesis and depth.

“Huntress reports that Evil Tokens phishing as a service operation drove a 1,380% increase in device code phishing attacks in early 2026”

“type D which is configuration file only and this tends to be silent. Now the agent leaves traces in fewer than 0.5% of commits”

The episode recycles well-worn cybersecurity tropes in the first segment (alert fatigue, tool sprawl, 'context is king') and the second segment is explicitly a curated summary of public news articles with minimal original editorial framing or contrarian analysis. Almost no first-principles thinking is present.

“The attacks don't stop. They really don't. Um, it may sound cliche but it's real”

“knowledge is power. Let's see everything that we can”

The sole 'guest' clip is from a Check Point internal employee (Michael Greenberg, exposure management team) delivering what reads as a product-marketing pitch for Check Point's own tooling; the host Phoneboy is a Check Point brand persona. No independent external practitioners or verifiable at-scale operators appear.

“why is Jacob's team working so hard? Why is R and D working so hard to develop uh, ah, security”

“there's also a white paper there that um, goes into more detail and also um, you can look into our AI red teaming tool”

The AI news segment includes several concrete data points with named sources (Evil Tokens, NordStellar, ARCsive, OA Labs, Huntress) and specific figures, and the agent-trace detection taxonomy is genuinely detailed. However, attribution is secondhand and the ransomware statistics in the first segment are vague and unsourced.

“a NORD Stellar report found a 39% rise in dark web discussion of deepfake as a service offerings compared with last year”

“a new ARCsive study detected AI coding agents traces across 180 million repositories”

There is no actual conversation in this episode: the first half is a replayed clip from a previous webinar and the second half is the host reading a pre-written newsletter summary aloud. No follow-up questions, no pushback, no dialogue of any kind occurs.

“The link to the full session is of course available in the show notes”

“let me know what you think by leaving a comment on the show post on Checkmates”