Épisodes

  • Adversarial Podcast S4E08 – Shai-Hulud worm strikes again, critical React vuln, CrowdStrike insider threat
    Dec 9 2025

    00:00 Intro

    02:33 Shai Hulud 2.0

    17:12 Max severity React vulnerability

    29:23 CrowdStrike catches insider feeding information to hackers

    46:24 Anthropic disruptes AI-orchestrated cyber campaign

    52:35 Uncertain economy takes effect on cyber teams

    Shai-Hulud 2.0 Aftermath: Trends, Victimology and Impact

    Researchers report that Shai-Hulud 2.0 is an ongoing npm supply-chain worm that has compromised hundreds of packages and tens of thousands of GitHub repositories and siphoned secrets through CI/CD pipelines.

    Critical React Server Components Vulnerability CVE-2025-55182

    React vulnerability React Server Components (RSC) — tracked as CVE-2025-55182 — is a critical (CVSS 10.0) flaw that allows unauthenticated attackers to execute arbitrary code on servers just by sending a crafted HTTP request to vulnerable packages.

    CrowdStrike catches insider feeding information to hackers

    CrowdStrike caught an insider who had secretly shared screenshots of internal systems with hackers linked to Scattered Lapsus$ Hunters — though the company says no breach of its infrastructure occurred and no customer data was compromised.

    Comcast's 2025 Cybersecurity Threat Report

    Comcast Business’s 2025 Cybersecurity Threat Report finds that over the 12-month period ending May 31, 2025 the company recorded 34.6 billion cyber events — including 4.7 billion phishing attempts, 9.7 billion “drive-by” compromise attacks, 44,000 DDoS attacks, and 19.5 billion resource-development activities.

    Disrupting the first reported AI-orchestrated cyber espionage campaign

    Anthropic reports disrupting what it assesses to be the first large-scale, AI-orchestrated cyber espionage campaign, in which a Chinese state-linked group jailbroke Claude Code to autonomously conduct reconnaissance, exploit vulnerabilities, and exfiltrate data across dozens of global targets with minimal human involvement.

    Uncertain Economy Takes Toll on Cybersecurity Teams

    Economic uncertainty has hit corporate cyber operations: Artico Search and IANS Research report that cybersecurity budgets rose just 4% in 2025 (a five-year low), hiring growth slowed to 7% (down from 12% in 2024), and many security-teams are grappling with tighter budgets, fewer hires, and slower wage growth.

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (Founder, http://githoundexplore.com/)

    Voir plus Voir moins
    1 h et 2 min
  • Adversarial Podcast S4E07 – The password is "Louvre", AI ransomware, Nevada stands up to ransomware
    Nov 11 2025

    00:00 Intro

    01:50 Louvre password

    08:54 Trump budget cuts

    20:35 Google AI threat report

    36:56 Nevada didn’t pay ransom

    48:25 Moved the needle

    58:38 L3Harris Trenchant boss stole exploits, sold to Russia

    62:00 Ransomware remediation firm employees go rogue

    63:40 Cybersecurity Is A Digital Identity Problem And We Must Deal With It

    The password for the Louvre’s video surveillance system was “Louvre”

    The Louvre Museum reportedly had a video-surveillance server password of simply “LOUVRE” as early as 2014..

    Trump budget cuts, agency gutting, leave Americans and economy at greater risk of being hacked, experts warn

    Budget cuts under Donald Trump’s administration are slashing funding and staff at key federal cybersecurity agencies like CISA, increasing the risk of U.S. vulnerability to cyberattacks.

    GTIG AI Threat Tracker: Advances in Threat Actor Usage of AI Tools | Google Cloud Blog

    Adversaries are now deploying AI-enabled malware (such as self-modifying code) and exploiting underground AI tool markets across the full attack lifecycle.

    Nevada didn’t pay ransom in statewide cyberattack, spent $1.5M on response

    The State of Nevada did not pay the ransom after a statewide cyberattack, opting instead to spend approximately $1.5 million on response efforts.

    How an ex-L3Harris boss stole and sold cyber exploits to Russia

    A former L3Harris division boss admitted to stealing eight zero-day exploits from network and selling them to a Russian cyber-tool broker.

    Chicago firm that resolves ransomware attacks had rogue workers carrying out their own hacks, FBI says

    A Chicago-based ransomware response firm is under indictment after employees allegedly conducted five ransomware attacks of their own.

    Cybersecurity Is A Digital Identity Problem And We Must Deal With It

    Cybersecurity failures increasingly stem from weak or mis-managed digital identities, and organizations must shift their focus from endpoints to identity-first strategies.

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (Founder, http://githoundexplore.com/)

    Voir plus Voir moins
    1 h et 14 min
  • Adversarial Podcast S4E06 – F5 Breach, AWS Outage, Risk Management vs. Security Engineering
    Oct 28 2025

    00:00 Intro

    00:50 AWS Outage

    20:48 F5 Breach

    41:06 Risk Management vs. Security Engineering

    58:19 Moving the Needle Part 3

    F5 Hack Blamed on China

    Chinese state-backed hackers allegedly breached U.S. cybersecurity firm F5, gaining year-long access to its systems and BIG-IP source code, prompting security fears and causing the company to warn of revenue impacts and falling shares.

    AWS Outage

    A race condition in Amazon DynamoDB’s DNS management system caused widespread outages across the US-EAST-1 region on October 19–20, 2025, disrupting DynamoDB, EC2, NLB, and multiple dependent AWS services until recovery was completed the next afternoon.

    The CISO Dilemma: Risk Management vs. Security Engineering

    This post argues that quantitative risk management (QRM) in cybersecurity is a deceptive comfort mechanism that lets executives rationalize insecurity, urging CISOs to reject financialized “risk buy-downs” and instead demand true security engineering and systemic architectural integrity.

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (Founder, http://githoundexplore.com/)

    Voir plus Voir moins
    1 h et 12 min
  • Adversarial Podcast S4E05 – Oracle Zero-Day, US cyber info sharing law expires, UK government guarantor for Jaguar attack
    Oct 14 2025

    00:00 Highlight

    03:44 Oracle E-Business Suite Zero-Day

    14:49 UK government to be guarantor for Jaguar Land Rover cyberattack

    25:54 "Moved the needle" Part 2

    48:18 12 Security Problems Practitioners Want Solved

    1:02:53 National Risk of Losing the CISA 2015 Act?

    Oracle E-Business Suite Zero-Day Exploited in Widespread Extortion Campaign

    Mandiant and Google Threat Intelligence Group uncovered a large-scale CL0P-linked extortion campaign exploiting a zero-day (CVE-2025-61882) in Oracle E-Business Suite to steal data from organizations before patches were released.

    https://cloud.google.com/blog/topics/threat-intelligence/oracle-ebusiness-suite-zero-day-exploitation

    UK government to be guarantor for Jaguar Land Rover loan as it recovers from cyberattack

    The UK government is guaranteeing a £1.5 billion loan to Jaguar Land Rover to support its recovery and supply chain after a major cyberattack forced the automaker to halt production earlier this month.

    https://therecord.media/jaguar-land-rover-loan-guarantor-cyberattack

    12 Security Problems Practitioners Want Solved

    Leen and Lockstep Ventures released a “Requests for Security Startups” report outlining twelve practitioner-driven problem areas—from preventative security and identity sprawl to AI-native assistants and continuous compliance—calling for builders to create practical, AI-powered, and workflow-integrated solutions that solve real security pain points.

    https://www.leen.dev/beyond-the-noise

    When Cyber Visibility Fades: The National Risk of Losing the CISA 2015 Act—and How Organizations Can Stay Secure Without It

    The expiration of the Cybersecurity Information Sharing Act of 2015 has reduced national cyber visibility and weakened public–private threat intelligence sharing, prompting experts to warn that organizations must strengthen internal risk management and collaboration to stay secure.

    https://www.carson-saint.com/when-cyber-visibility-fades-the-national-risk-of-losing-the-cisa-2015-act-and-how-organizations-can-stay-secure-without-it

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (Founder, http://githoundexplore.com/)

    Voir plus Voir moins
    1 h et 11 min
  • Adversarial Podcast S4E04 – "Moving the needle" awards, effect of H-1B changes on cyber industry, Salesloft aftermath
    Sep 30 2025

    00:00 Highlight

    00:43 Intro

    06:40 "Moved the needle" awards

    37:05 Scattered Lapsus$ and Jaguar Hack

    44:39 One Token to Rule Them All - Entra pwned

    1:02:21 H-1B visa changes and their effect on the cyber industry

    Scattered Lapsus$ and Jaguar Hack

    Jaguar Land Rover has extended its production pause until October after a cyberattack crippled its IT systems. The company is struggling to recover operations at Range Rover plants.

    https://www.wsj.com/business/jaguar-land-rover-extends-production-pause-until-october-following-cyberattack-0e39b7e8

    One Token to Rule Them All

    A deep dive into how attackers can obtain Global Admin across all Entra ID tenants using Actor tokens — the mechanics, prerequisites, and mitigation strategies.

    https://dirkjanm.io/obtaining-global-admin-in-every-entra-id-tenant-with-actor-tokens/

    What to Know About Changes to the H-1B Visa Program

    The U.S. is proposing major H-1B visa changes, including a $100,000 annual fee per visa starting in 2026, a move aimed at prioritizing higher-wage hires but likely to hit startups and global tech talent hard.

    https://www.wsj.com/us-news/h1b-visa-changes-explained-45b818e9?mod=djemCybersecruityPro

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (Founder, http://githoundexplore.com/)

    Voir plus Voir moins
    1 h et 19 min
  • Adversarial Podcast S4E03 – Fumbled NPM Attack, Entering the AI Browser Market, Salesloft breach
    Sep 16 2025

    00:00 Intro

    03:10 NPM supply chain attack leaves attackers empty handed

    24:44 Why is Atlassian buying a browser company?

    37:20 Apple's new Memory Integrity Enforcement

    52:56 Salesloft breach leads to downstream hacks

    Hackers left empty-handed after massive NPM supply-chain attack

    Hackers briefly compromised popular NPM packages like chalk and debug-js, infecting ~10% of cloud environments, but despite the massive supply-chain reach they only netted about $600 in stolen cryptocurrency.

    https://www.bleepingcomputer.com/news/security/hackers-left-empty-handed-after-massive-npm-supply-chain-attack/

    Why is Atlassian Buying a Browser Company?

    Atlassian is buying The Browser Company (makers of Arc and Dia) for $610M to gain control of the browser channel, secure its AI agent (Rovo) distribution, and enter the emerging “enterprise browser” market, even though success is uncertain against Google and Microsoft.

    https://nextword.substack.com/p/why-is-atlassian-buying-a-browser

    Memory Integrity Enforcement: A complete vision for memory safety in Apple devices

    Apple’s new Memory Integrity Enforcement (MIE) brings always-on hardware-software memory safety to iPhone 17, making advanced spyware exploits far harder.

    https://security.apple.com/blog/memory-integrity-enforcement/

    Salesloft breached to steal OAuth tokens for Salesforce data-theft attacks

    Hackers exploited Salesloft’s Drift–Salesforce integration to steal OAuth tokens and exfiltrate sensitive Salesforce data, tracked as UNC6395.

    https://www.bleepingcomputer.com/news/security/salesloft-breached-to-steal-oauth-tokens-for-salesforce-data-theft-attacks/

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (Founder, https://githoundexplore.com)

    Voir plus Voir moins
    1 h et 9 min
  • Adversarial Podcast S4E02 - Cyber acquisitions and raises, 95% of GenAI pilots failing, Zelle's alleged security lapses
    Sep 4 2025

    00:00 Introduction & BlackHat

    02:06 Cybersecurity in Schools

    18:53 Black Hat Conference Highlights

    34:02 New York sues Zelle

    44:48 Trends in Cybersecurity Mergers and Acquisitions

    1:02:44 95% of generative AI pilots at companies are failing

    1:08:53 Prompt injection with poisoned calendar invites

    DARPA announces $4 million winner of AI code review competition at DEF CON

    DARPA announced Team Atlanta as the winner of its two-year competition among researchers to create the best artificial intelligence systems that can find and fix vulnerabilities.

    Attorney General James Sues Company Behind Zelle for Enabling Widespread Fraud

    New York today sued Early Warning Services, a company owned and controlled by a group of the largest banks in the United States that was tasked with developing and operating the electronic payment platform Zelle, for failing to protect its users from massive amounts of fraud.

    Cyber Acquisitions

    • Palo Alto / CyberArk
    • CrowdStrike / Onum
    • Okta / Axiom
    • Armis raises millions at $5B valuation

    MIT report: 95% of generative AI pilots at companies are failing

    A recent MIT‑commissioned study—highlighted in Fortune on August 18, 2025—reveals that approximately 95% of generative AI pilot programs at companies failed to deliver any measurable return on investment or financial uplift. The core issue appears to be not the AI itself, but poor integration into existing workflows and misaligned use cases, with only about 5% of pilots achieving rapid revenue growth by focusing sharply on specific pain points.

    Hackers Hijacked Google’s Gemini AI With a Poisoned Calendar Invite to Take Over a Smart Home

    Security researchers demonstrated that a poisoned Google Calendar invite could indirectly prompt-inject Google’s Gemini, causing it to control smart-home devices.

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (https://tillsongalloway.com)

    Voir plus Voir moins
    1 h et 16 min
  • Adversarial Podcast S4E01 - Trump's AI Action Plan, Chip Security Act, receiving gifts from vendors
    Jul 30 2025

    00:00 Introduction & BlackHat

    03:14 AI Action Plan Overview

    13:30 Chip Security Act

    20:48 Government led AI-ISAC?

    23:16 UK government considering banning public sector ransomware payments

    28:14 Microsoft probing if Chinese hackers learned SharePoint flaws through alert

    42:07 Ethics in Vendor Relationships – Gifts for meetings

    America's AI Action Plan

    “America’s AI Action Plan,” released by the Trump administration, outlines a roadmap with over 90 federal actions across three pillars—accelerating AI innovation, building U.S. AI infrastructure, and asserting international AI leadership through exports and technology alliances.

    The Chip Security Act: A Bipartisan Solution to Chip Smuggling

    The Chip Security Act, introduced by U.S. lawmakers, mandates that export‑controlled AI chip makers (like NVIDIA) embed on‑chip location‑verification mechanisms to ensure devices go only where they’re authorized—aiming to deter smuggling (especially to China) without deploying intrusive GPS or kill switches.

    Why a Government-Led AI-ISAC is a Missed Opportunity

    Errol Weiss argues that an AI‑ISAC led by the U.S. government, as proposed in the July 2025 White House AI Action Plan, represents a missed opportunity, because government-led initiatives tend to be bureaucratic, slow, less innovative, struggle to win private-sector trust and buy‑in, risk duplicating existing ISAC efforts, and may be perceived as politically biased—undermining effective, rapid, cross-industry intelligence sharing

    UK plans to ban public sector bodies from paying ransom to cyber criminals

    The UK government is set to ban public sector bodies and operators of critical national infrastructure from paying ransom demands to cyber criminals, as part of a wider package also mandating mandatory reporting for other organisations planning to pay, aimed at dismantling the ransomware business model and protecting essential services from dangerous disruptions.

    Microsoft probing if Chinese hackers learned SharePoint flaws through alert, Bloomberg News reports

    Microsoft is investigating whether a leak from its Microsoft Active Protections Program (MAPP)—which provides early vulnerability alerts to security partners—may have enabled Chinese-aligned hackers (Linen Typhoon, Violet Typhoon, and Storm-2603) to exploit critical zero‑day flaws in on-premises SharePoint servers before Microsoft fully patched the software, fueling a global espionage and ransomware campaign.

    Hosts:

    Jerry Perullo (Founder, https://adversarial.com/)

    Sounil Yu (Founder, https://www.knostic.ai/)

    Mario Duarte (Founder, stealth startup)

    Producer: Tillson Galloway (https://tillsongalloway.com)

    Voir plus Voir moins
    52 min