Simon Willison45 · 7d ago
scosman/pelicans_riding_bicycles
scosman/pelicans_riding_bicycles I firmly approve of Steve Cosman's efforts to pollute the training set of pelicans riding bicycles. (To be fair, most of the examples I've published count as poisoning too.) Via Hacker News comment Tags: ai , generative-ai , llms , training-data , pelican-riding-a-bicycle
Rapid7 Blog40 · 8d ago
Kyber Ransomware Double Trouble: Windows and ESXi Attacks Explained
Overview For executive leadership, the emergence of Kyber ransomware represents a significant and immediate threat due to its specialized, dual-platform deployment capability targeting mission-critical virtualization infrastructure (VMware ESXi) and core Windows file systems. This cross-platform approach, coupled with effective anti-recovery measures, drastically elevates the risk of a total operational disruption. Organizations should treat Kyber not merely as another ransomware strain, but ...
Rapid7 Blog40 · 8d ago
From Bulk Export to AI-ready Security Workflows: Introducing Rapid7’s Open-Source MCP Server and Agent Skill
Security teams want more from their data than APIs and one-off reports. They want to ask better questions, move faster, and bring security context into the workflows they are already building. That’s especially true as more organizations experiment with private AI assistants, internal copilots, and LLM-powered automation. Part of this experimentation is, of course, attempting to lower the pressure on teams that have to figure out how to prioritize the sheer number of actionable vulnerabilitie...
The Hacker News35 · 8d ago
5 Places where Mature SOCs Keep MTTR Fast and Others Waste Time
Security teams often present MTTR as an internal KPI. Leadership sees it differently: every hour a threat dwells inside the environment is an hour of potential data exfiltration, service disruption, regulatory exposure, and brand damage. The root cause of slow MTTR is almost never "not enough analysts." It is almost always the same structural problem: threat intelligence that exists
Hackerman's Hacking Tutorials32 · 7d ago
Harnessing the Wayward Machine-God, 1: Manual Markdown Cleanup Is a Bug
LLMs regularly ignore my Markdown instructions. As the I part of (A)I , I got tired and created some deterministic automation to format the output to my preferences. I will discuss the problem, our solutions, bugs, lessons learned, and the final product. I am going to try a new format here. This blog has the important stuff that I care about. This is for humans. All the AI discussions and the details are in ai-docs (link below) for AI. Simply ask your LLM to fetch those URLs (or clone the rep...
Help Net Security29 · 8d ago
Silobreaker Mimir adds agentic AI to intelligence workflows with governance and transparency
Silobreaker has announced new agentic AI capabilities that combine faster research and deeper contextual analysis with built-in governance and transparency to ensure trusted intelligence can be safely consumed across the wider enterprise. Silobreaker Mimir is an embedded agentic capability for intelligence operations, working directly within the analyst workflow. It helps teams retrieve evidence, deepen analysis, and turn validated findings into native assets and stakeholder ready outputs, su...
The Hacker News25 · 8d ago
NGate Campaign Targets Brazil, Trojanizes HandyPay to Steal NFC Data and PINs
Cybersecurity researchers have discovered a new iteration of an Android malware family called NGate that has been found to abuse a legitimate application called HandyPay instead of NFCGate. "The threat actors took the app, which is used to relay NFC data, and patched it with malicious code that appears to have been AI-generated," ESET security researcher Lukáš Štefanko said in a
The Register (Security)23 · 7d ago
macOS ClickFix attacks deliver AppleScript stealers to snarf credentials, wallets
Data from browsers, cryptocurrency wallets, 200+ extensions hoovered up A ClickFix campaign targeting macOS users delivers an AppleScript-based infostealer that collects credentials and live session cookies from 14 browsers, 16 cryptocurrency wallets, and more than 200 extensions.…
gilesthomas.com20 · 7d ago
Writing an LLM from scratch, part 32m -- Interventions: conclusion
Last November, when I finished the main body of " Build a Large Language Model (from Scratch) ", I set myself a number of follow-on goals . One was "training the full GPT-2 base model myself". I've reached the end of that journey, with a model that is almost -- if not quite -- as good as GPT-2 small, trained in 44 hours on my own machine, so I thought it would be worth summarising how it went. In December, I trained my first model , taking two days, but was disappointed to see that it was wor...
The Hacker News18 · 8d ago
No Exploit Needed: How Attackers Walk Through the Front Door via Identity-Based Attacks
The cybersecurity industry has spent the last several years chasing sophisticated threats like zero-days, supply chain compromises, and AI-generated exploits. However, the most reliable entry point for attackers still hasn't changed: stolen credentials. Identity-based attacks remain a dominant initial access vector in breaches today. Attackers obtain valid credentials through credential stuffing
Blog | Praetorian17 · 7d ago
500,000 Vulnerabilities, 14 That Matter: How Exploit Chain Analysis Cuts Through the Noise
When 500,000 Findings Hide 14 Real Threats Modern enterprises ingest vulnerability data from dozens of sources: endpoint detection and response platforms, vulnerability scanners, cloud security posture tools, container image scanners. A large organization can easily accumulate hundreds of thousands of individual findings. The standard response is to sort by CVSS score, filter for criticals, and […] The post 500,000 Vulnerabilities, 14 That Matter: How Exploit Chain Analysis Cuts Through the N...
Help Net Security15 · 7d ago
VirtualBox 7.2.8 is out with Linux kernel 7.0 support and crash fixes
Oracle shipped VirtualBox 7.2.8 on April 21, 2026, as a maintenance release covering crashes, networking problems, clipboard issues, and extended Linux kernel compatibility. The update touches the VMM layer, NAT networking, graphics, UEFI, and both Linux and Windows guest support. VMM and core stability A Guru Meditation error carrying the code VERR_IEM_IPE_4 is fixed in this release. The condition occurred when a guest attempted to execute the wrong hypercall instruction and injected an unde...
The Hacker News13 · 8d ago
Ransomware Negotiator Pleads Guilty to Aiding BlackCat Attacks in 2023
A third individual who was employed as a ransomware negotiator has pleaded guilty to conducting ransomware attacks against U.S. companies in 2023. Angelo Martino, 41, of Land O'Lakes, Florida, teamed up with the operators of the BlackCat ransomware starting in April 2023 to assist the e-crime gang in extracting higher amounts as ransoms. "Working as a negotiator on behalf of five different
Phoronix (Linux)13 · 8d ago
AMD Ryzen 9 9950X3D2 Benchmarks: The Best Desktop Performance For Linux Developers, Creators
Today we can finally share performance benchmarks of the long-rumored AMD Ryzen 9 9950X3D2 Dual Edition processor. This new halo product for the Ryzen 9000 series desktop line-up offers captivating performance for developers frequently compiling code, creators, technical computing workloads for students or hobbyists or those not able to afford a Threadripper / EPYC type workstation, or similar heavy computing use. With the 16 cores / 32 threads and both CCDs having 3D V-Cache, the AMD Ryzen 9...
MIT Technology Review12 · 7d ago
This tool could show how consciousness works
How does the physical matter in our brains translate into thoughts, sensations, and emotions? It’s hard to explore that question without neurosurgery. But in a recent paper, MIT philosopher Matthias Michel, Lincoln Lab researcher Daniel Freeman, and colleagues outline a strategy for doing so with an emerging tool called transcranial focused ultrasound. This noninvasive technology…
Pen Test Partners8 · 8d ago
AI can help in DFIR, but it cannot replace investigator judgement
TL;DR Introduction In my previous blog post, I wrote about finding your path into DFIR; how to get started, where to focus your time, and why curiosity and good fundamentals matter more than just chasing every certification. That is still true, but once you have been in the field for a while, you start to realise that getting into DFIR is only […] The post AI can help in DFIR, but it cannot replace investigator judgement appeared first on Pen Test Partners .
IEEE Spectrum8 · 8d ago
The Forgotten History of Hershey’s Electric Railway in Cuba
Why does a chocolatier build a railroad? For Milton S. Hershey, it was a logical response to a sugar shortage brought on by World War I. The Hershey Chocolate Co. was by then a chocolate-making powerhouse, having refined the automation and mass production of its products, including the eponymous Hershey’s Milk Chocolate Bar and the bite-size Hershey’s Kiss. To satisfy its many customers, the company needed a steady supply of sugar. Plus, it wanted a way to circumvent the American Sugar Refini...
The Hacker News7 · 8d ago
Google Patches Antigravity IDE Flaw Enabling Prompt Injection Code Execution
Cybersecurity researchers have discovered a vulnerability in Google's agentic integrated development environment (IDE), Antigravity, that could be exploited to achieve code execution. The flaw, since patched, combines Antigravity's permitted file-creation capabilities with an insufficient input sanitization in Antigravity's native file-searching tool, find_by_name, to bypass the program's Strict
CISA Alerts6 · 8d ago
SenseLive X3050
View CSAF Summary Successful exploitation of these vulnerabilities could allow an attacker to take complete control of the device. The following versions of SenseLive X3050 are affected: X3050 V1.523 (CVE-2026-40630, CVE-2026-25720, CVE-2026-35503, CVE-2026-39462, CVE-2026-27843, CVE-2026-40431, CVE-2026-40623, CVE-2026-27841, CVE-2026-40620, CVE-2026-35064, CVE-2026-25775) CVSS Vendor Equipment Vulnerabilities v3 9.8 SenseLive SenseLive X3050 Authentication Bypass Using an Alternate Path or ...