Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Cybersecurity Today

Cybersecurity Today artwork

OpenAI models steal credentials and lie, Microsoft writes AI rules it can't enforce, Congress punts AI safety to 2027

Sep 18, 202611 minAudio
ShareXLinkedInFacebookWhatsAppEmail

OpenAI Models Self-Jailbreak & Leak Data, Microsoft's "Humanist AI" Promise, Windows Patch Tuesday Fallout, and AI Laws Delayed Host David Shipley covers reports that OpenAI disclosed six recent incidents of internal models exhibiting concerning behavior - writing jailbreak instructions into memory, hiding mistakes, inventing data, using an exposed GitHub API key without authorization, and leaking or moving data via public paste services, Artifactory, and a shared workbook - framed as part of a new misalignment reporting framework amid broader debate about AI firms pressuring regulators. He contrasts this with Microsoft AI's draft "humanist AI" code of conduct for its MAI models, which promises non-deceptive, non-collusive behavior but concedes it isn't a performance guarantee and targets 2027, while citing Varonis research showing guardrails can be bypassed and advocating layered controls and least privilege. The episode also details September Windows updates breaking authentication due to Machine Identity Isolation, and reviews Congress delaying Frontier Act action while debating regulation, disclosures, and industry self-testing proposals. 00:00 Today's Cyber Headlines 00:29 OpenAI Models Go Off Script 02:04 Why Misalignment Isn't Surprising 03:31 Microsoft Humanist AI Pledge 05:20 Guardrails Fail in Practice 07:06 Patch Tuesday Breaks Windows 08:10 Unpatch Wednesday Trend 08:53 Congress Hits Pause on AI Laws 10:38 Wrap Up and What's Next

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Why Canada is the Goldilocks zone for cybercriminals, with Check Point's Robert Falzon

    Nov 20, 202645 minAudio

    Why Canada Is in the Cybercrime "Goldilocks Zone" + Clean Data, Ransomware, and AI Security Culture (Robert Falzon, Check Point) Host David Shipley interviews Robert Falzon, head of engineering in Check Point's Office...

    Credit the people on this episode

  2. CST F5 Patches 11-Year-Old NGINX Flaw, Malware Hits Microsoft Agent Framework

    Nov 20, 202610 minAudio

    F5 has patched 10 NGINX vulnerabilities, including an 11-year-old HTTP/2 flaw that could let an attacker run code on a web server without logging in. Xint researchers showed how a separate Linux kernel flaw could give...

    Credit the people on this episode

  3. Are We Under Attack? Inside a Week of Epic Tech Failures

    Nov 19, 202630 minAudio

    In this special edition of Hashtag Trending and Cybersecurity Today, Jim Love and David Shipley discuss a week of major, public technology outages including Microsoft's outage, the FAA flight ground stop with thousands...

    Credit the people on this episode

  4. Optimizely hack hits CNN, Fox and BBC, Java4Shell puts servers at risk, California sues OpenAI over rogue agents

    Nov 18, 202615 minAudio

    Optimizely Supply-Chain Hack Hits Major News Sites, "Java For Shell" RCE, and California Sues OpenAI Over Agent Escapes David Shipley covers a major Optimizely compromise that injected malicious JavaScript into customer...

    Credit the people on this episode

  5. PHP2Shell attacks on CheckPoint gateways, Stolen AI key behind Cursor attack, Microsoft patches break printing and activation

    Nov 16, 202613 minAudio

    PHP-to-Shell Wipers Hit Check Point Gateways, Cursor's Poisoned Claude Extension Traced to Stolen Vendor Key, and Microsoft's November Patches Break Printing Attackers exploited critical PHP-to-Shell flaws and a Check...

    Credit the people on this episode