1574 matches found
Meta Resumes E.U. AI Training Using Public User Data After Regulator Approval
Meta has announced that it will begin to train its artificial intelligence AI models using public data shared by adults across its platforms in the European Union, nearly a year after it paused its efforts due to data protection concerns from Irish regulators. "This training will better support...
X-Teaming: Multi-Turn Jailbreaks and Defenses with Adaptive Multi-Agents
Multi-turn interactions with language models LMs pose critical safety risks, as harmful intent can be strategically spread across exchanges. Yet, the vast majority of prior work has focused on single-turn safety, while adaptability and diversity remain among the key challenges of multi-turn...
RAID: an In-Training Defense against Attribute Inference Attacks in Recommender Systems
In various networks and mobile applications, users are highly susceptible to attribute inference attacks, with particularly prevalent occurrences in recommender systems. Attackers exploit partially exposed user profiles in recommendation models, such as user embeddings, to infer private attribute...
How to Enhance Downstream Adversarial Robustness (Almost) without Touching the Pre-Trained Foundation Model?
With the rise of powerful foundation models, a pre-training-fine-tuning paradigm becomes increasingly popular these days: A foundation model is pre-trained using a huge amount of data from various sources, and then the downstream users only need to fine-tune and adapt it to specific downstream...
Meta slurps up EU user data for AI training
European Facebook users have so far avoided having their public posts used to train parent company Meta's AI model. That's about to change, the company has warned. In a blog post today, it said that EU residents' data was fair game and it would be slurping up public posts for training soon...
Explore how to secure AI by attending our Learn Live Series
As organizations develop, use, and increasingly rely on AI applications, they must address new and amplified security risks. Are you prepared to secure your environment for AI adoption? How about identifying threats to your AI and safeguarding data? Register to attend one or all our Learn Live...
IBM: Information disclosure on IBM training service endpoint
The IBM training service endpoint had an information disclosure vulnerability that was reported to IBM, analyzed, and remediated. The vulnerability was discovered and reported by an external researcher...
CVE-2025-3410
A vulnerability classified as critical was found in mymagicpower AIAS 20250308. This vulnerability affects unknown code of the file trainingplatform/train-platform/src/main/java/top/aias/training/controller/LocalStorageController.java. The manipulation of the argument File leads to unrestricted...
PT-2025-15357 · Unknown · Mymagicpower Aias
Name of the Vulnerable Software and Affected Versions: mymagicpower AIAS 20250308 Description: A critical issue was found in mymagicpower AIAS, affecting an unknown function of the file 2 training platform/train-platform/src/main/java/top/aias/training/controller/InferController.java. The...
G2 Names INE 2025 Cybersecurity Training Leader
Cary, North Carolina, 27th March 2025, CyberNewsWire...
Rapid7 Earns 5-Star Rating in the 2025 CRN® Partner Program Guide
Rapid7 has been honored by CRN®, a brand of The Channel Company, with a 5-Star Award in the 2025 CRN Partner Program Guide. This annual guide is an essential resource for solution providers seeking vendor partner programs that match their business goals and deliver high partner value. Recognition...
Cyber Guardians: INE Security Champions Cybersecurity Training During National Physicians Week 2025
Cary, NC, 24th March 2025, CyberNewsWire...
CVE-2024-9052
A flaw was found in the vLLM distributed training API. This vulnerability allows remote code execution via unsafe deserialization, which uses pickle.loads without sanitization...
My Writings Are in the LibGen AI Training Corpus
The Atlantic has a search tool that allows you to search for specific works in the "LibGen" database of copyrighted works that Meta used to train its AI models. The rest of the article is behind a paywall, but not the search tool. It’s impossible to know exactly which parts of LibGen Meta used to...
CVE-2024-9052
Rejected reason: This CVE ID has been rejected or withdrawn by its CVE Numbering Authority...
CVE-2024-9052
Summary: CVE-2024-9052 relates to a deserialization flaw in the vLLM project’s distributed training API. The issue enables remote code execution via unsafe deserialization of object bytes using pickle.loads() without sanitization, specifically within the vllm.distributed.GroupCoordinator.recv_obj...
The vulnerability of the Iced training platform’s classes allows a hacker to execute arbitrary code.
The vulnerability of the Iced training platform for machine learning H2O is related to deficiencies in the deserialization mechanism. Exploiting this vulnerability allows a remote attacker to execute arbitrary code...