privacy-filter

OpenAI Privacy Filter OpenAI Privacy Filter is a bidirectional token-classification model for personally identifiable information PII detection and masking in text. It is intended for high-throughput data sanitization workflows where teams need a model that they can run on-premises that is fast,...

OpenAI Launches GPT-5.6-Cyber with Reduced Safeguards for Exploit Development

OpenAI on Monday unveiled a new cybersecurity-focused model called GPT‑5.6‑Cyber that it said is focused on vulnerability research, penetration testing, and incident response. "Built on GPT‑5.6 Sol, it is trained to improve capabilities on several specialized cybersecurity tasks e.g., finding...

Mako: A Self-Evolving Agentic Operating System (SE-AOS) for Autonomous Web Exploitation

We introduce the Self-Evolving Agentic Operating System SE-AOS: a new class of AI agent that treats exploit capability as a mutable, versioned kernel it extends at runtime, observing its own failures, synthesising new capabilities, proving them against a live target, and hot-loading them back int...

Large Language Models (LLMs) and Generative AI in Cybersecurity and Privacy: A Survey of Dual-Use Risks, AI-Generated Malware, Explainability, and Defensive Strategies

Large Language Models LLMs and generative AI GenAI systems, such as ChatGPT, Claude, Gemini, LLaMA, Copilot, Stable Diffusion by OpenAI, Anthropic, Google, Meta, Microsoft, Stability AI, respectively, are revolutionizing cybersecurity, enabling both automated defense and sophisticated attacks...

Ransomware Groups Exploit Legit IT Tools to Bypass Antivirus

New research from Seqrite explains the 'dual-use dilemma,' where ransomware attackers repurpose legitimate IT tools like IOBit Unlocker…...

CVE-2026-32897: Undefined Security Weakness

OpenClaw versions prior to 2026.2.22 reuse gateway.auth.token as a fallback hash secret for owner-ID prompt obfuscation when commands.ownerDisplay is set to hash and commands.ownerDisplaySecret is unset, creating dual-use of authentication secrets across security domains. Attackers with access to...

PT-2026-25487

Vulnerability OpenClaw reused gateway.auth.token and gateway.remote.token as a fallback hash secret for owner-ID prompt obfuscation when commands.ownerDisplay=hash and commands.ownerDisplaySecret was unset. This created secret dual-use between gateway authentication and prompt metadata hashing...

Russian Hackers Target Ukrainian Organizations Using Stealthy Living-Off-the-Land Tactics

Organizations in Ukraine have been targeted by threat actors of Russian origin with an aim to siphon sensitive data and maintain persistent access to compromised networks. The activity, according to a new report from the Symantec and Carbon Black Threat Hunter Team, targeted a large business...

On the Ethics of Using LLMs for Offensive Security

Large Language Models LLMs have rapidly evolved over the past few years and are currently evaluated for their efficacy within the domain of offensive cyber-security. While initial forays showcase the potential of LLMs to enhance security research, they also raise critical ethical concerns regardi...

Using LLMs to Create Bioweapons

Im not sure there are good ways to build guardrails to prevent this sort of thing: There is growing concern regarding the potential misuse of molecular machine learning models for harmful purposes. Specifically, the dual-use application of models for predicting cytotoxicity18 to create new poison...

Talos Takes 126: Year in Review - Threat Landscape Edition

Were back with another year in review focused episode. This time the focus will be the threat landscape generally and Ill be joined by threat researcher Caitlin Huey. In this episode well discuss what we found in the last year, with a focus on the general threat landscape. Well spend time...

Quarterly Report: Incident Response Trends in Q3 2022

Ransomware and pre-ransomware engagements make up 40 percent of threats seen this quarter By Caitlin Huey. For the first time since compiling these reports, Cisco Talos Incident Response saw an equal number of ransomware and pre-ransomware engagements, making up nearly 40 percent of threats this...

GitHub Updates Policy to Remove Exploit Code When Used in Active Attacks

Code-hosting platform GitHub Friday officially announced a series of updates to the site's policies that delve into how the company deals with malware and exploit code uploaded to its service. "We explicitly permit dual-use security technologies and content related to research into vulnerabilitie...

Kegtap, Singlemalt, Winekey Malware Serve Up Ransomware to Hospitals

The boozy names might sound like the kind of thing conjured up in a frat-house common room, but malware families Kegtap, Singlemalt and Winekey are being used to gain initial network access in potentially lethal ransomware attacks on healthcare organizations in the midst of a global pandemic,...

Fileless Malware Tops Critical Endpoint Threats for 1H 2020

In the first half of 2020, the most common critical-severity cybersecurity threat to endpoints was fileless malware, according to a recent analysis of telemetry data from Cisco. Fileless threats consist of malicious code that runs in memory after initial infection, instead of files being stored o...

WastedLocker Goes "Big-Game Hunting" in 2020

By Ben Baker, Edmund Brumaghin, JJ Cummings and Arnaud Zobec. Threat summary After initially compromising corporate networks, the attacker behind WastedLocker performs privilege escalation and lateral movement prior to activating ransomware and demanding ransom payment.The use of "dual-use" tools...

Mozilla SeaMonkey DoS

The Mozilla SeaMonkey browser contains a vulnerability related to errors in the implementation of a certain type of key verification mechanism in the cryptoGenerateCRMFRequest method. Exploiting this vulnerability allows malicious actors to induce a service failure abnormal termination of the...

Firefox DoS

Mozilla Firefox browser contains a vulnerability related to errors in the implementation of a certain type of key verification mechanism in the crypto.generateCRMFRequest method. Exploiting this vulnerability allows malicious actors, operating remotely, to cause a service failure abnormal...

Proposed U.S. Wassenaar Rules on Intrusion Software

Two things worth noting from yesterday’s unveiling of the Bureau of Industry and Security’s proposed Wassenaar rules for the U.S. that weren’t so overt: a The U.S. generally leads the way in implementing Wassenaar changes, and this time it’s been beaten by the EU by almost 18 months; and b reques...

Mozilla多个产品crypto.generateCRMFRequest拒绝服务漏洞

CVE ID:CVE-2014-1498 Mozilla Firefox/SeaMonkey是Mozilla所发布的WEB浏览器/新闻组客户端。 Mozilla多个产品在生成ec-dual-use请求时crypto.generateCRFMRequest方法没有正确校验KeyParams参数的键值类型，允许攻击者利用漏洞进行拒绝服务攻击，使应用程序崩溃。 0 Mozilla Firefox 27 Mozilla Seamonkey 2.24 Mozilla Firefox 28，Seamonkey 2.25已经修复该漏洞，建议用户下载更新： http://www.mozilla.org...