HackerOne’s Post

Hacking demos in AI can feel surprisingly easy. You grab some open datasets, fine-tune a well-known model, and boom—everything looks great! Your team sees a few accurately predicted examples, and suddenly it feels like you’ve won the AI war. Everyone's motivated and ready to celebrate. But then you ask the critical question: "Did we check accuracy across the entire dataset and all scenarios?" To your surprise, the answer reveals the war isn’t over—that initial success was just an illusion. The real challenge lies beyond the sample data, where true robustness and accuracy are tested. #confusionmatrix #recall #AIdeveloper #labeleddata #dataprep

Communication between AI projects and bug bounty hunters is crucial for ensuring smooth collaboration and effective vulnerability reporting. Here, Robert Morel explains how this works through our platform. https://lnkd.in/e-gjfweV

Don't miss out on "Graph Hunter" Episode 03 - "Showcase of Shadows"! Dive into how LLMs mingle with Knowledge Graphs for better fraud busting. Discover how AI is changing the fraud game. Secure your seat now - https://okt.to/x5B2XQ #ArangoDB #cogniware #machinelearning #DataSecurity #DataProtection #UserFriendly #GenerativeAI #KnowledgeGraphs #llms #CrimeFighting #webinar

Last day for #RSAC, let me know if you’re around! Taking in the morning session with OWASP® Foundation to better understand and defeat the top 10 emerging threats facing generative AI technologies (OWASP Top 10 For Large Language Model Applications)

Want to know what the experts at Palo Alto Networks think are the most concerning #AdversarialAI attacks? Tune into Threat Vector tomorrow to hear my conversation with AI experts William Hewlett and Tony H. on Threat Vector https://lnkd.in/dDwjBQKc.

It's a concern raised by many across the 911 community that the frequency of #swatting incidents are increasing. Some 911 centers are receiving multiple swatting calls every. single. day. The trend is alarming. Recent swatting incidents are plentiful and include the Alabama police officer arrested for swatting another police department ("he thought it was funny"), a swatting incident at the North Pole, and just yesterday, staff across Atlanta Public Schools received threatening emails that the FBI classified as a swatting attempt. Since we still lack a federal anti-swatting statute, states are taking matters into their own hands with Georgia and Utah to be the latest in over a dozen states to pass stricter legislation. But as a colleague recently pointed out that penalties, though necessary, won’t deter crime. For that reason, Mission Critical Partners is developing a "master class" to help the 911 community identify and mitigate swatting incidents. This class will cover topics of interest for all 911 professionals, including telecommunicators, supervisors, technologists, policy makers, and responders. For more details about this upcoming class, click here: https://lnkd.in/erHFJJfx

Day 1 of our 2024 Hack-ai-thon brought together some brilliant minds eager to innovate and tackle real-world challenges using AI. Check out some of the participants as they prepared to jump into the tracks. It will be a week of innovation and creativity. Hackers, may the code be with you all! #Nerdma #ThinkBeyond #Hackathon #ArtificialIntelligence

Are you aware that #hackers & #researchers are making money finding #bias in #artificialintelligence models? Congratulation to Keith Hoodlet on winning the first ever AI Bias Bounty program run by ConductorAI and Bugcrowd with DoD Chief Digital and Artificial Intelligence Office. The United States Department of Defense is utilizing #AI and #machinelearning models in incredible ways as a #forcemultiplier helping our #sailors, #soliders, #airmen, #marines, and #guardians become more productive, efficient, and lethal. To quote Chief of Naval Operations, Lisa Franchetti's priorities via her publishing of "America's Warfighting Navy" she outlines three priorities 1. Warfighting: Deliver Decisive Combat Power. 2. Warfighters: Strengthen the Navy Team. 3. Foundation: Build Trust, Align Resources, Be Ready. The UNITED STATES DEPARTMENT OF THE NAVY, US Army, United States Air Force, United States Space Force, United States Marine Corps, U.S. Coast Guard need to ensure they are utilizing and incorporating AI in the most responsible, efficient, and trusted ways. The DoD Chief Digital and Artificial Intelligence Office's Responsible AI toolkit (https://lnkd.in/edZNYxtc) should be one of the first considerations as they seek to deploy AI throughout the force. #dareando #informationsecurity #penetrationtesting #bugbounty #biasbounty #military

people ask me - they say "dex how can I win the AI Tinkerers Humans-in-the-Loop hackathon this weekend NOVEMBER 2-3" lemme tell ya, with all these top-tier hackers in one of the best cities for AI, it ain't gonna be easy my advice: - build something novel - build something useful - build something that works - stay true to the theme* *and if you wanna really crush human-in-the-loop AI systems, HumanLayer (YC F24) has you covered 🙂 #ai #aiAgents #humanInTheLoop

The use of AI has brought about a new set of challenges, as the attack surface of existing systems has increased beyond traditional cyber-attacks. MITRE has developed a new tool to help raise awareness of these unique and evolving vulnerabilities associated with LLM. MITRE ATLAS™ ( https://atlas.mitre.org/ ) is a globally accessible, living knowledge base of adversary tactics and techniques based on real-world attack observations and realistic demonstrations from AI red teams and security groups. ATLAS is modeled after the MITRE ATT&CK® framework and its tactics, techniques, and procedures (TTPs) are complementary to those in ATT&CK. #mitre #ai #techniques #attack