AI that talks to itself learns faster and smarterArtificial Intelligence News — ScienceDaily AI may learn better when it’s allowed to talk to itself. Researchers showed that internal “mumbling,” combined with short-term memory, helps AI adapt to new tasks, switch goals, and handle complex challenges more easily. This approach boosts learning efficiency while using far less training data. It could pave the way for more flexible, human-like AI systems.
AI may learn better when it’s allowed to talk to itself. Researchers showed that internal “mumbling,” combined with short-term memory, helps AI adapt to new tasks, switch goals, and handle complex challenges more easily. This approach boosts learning efficiency while using far less training data. It could pave the way for more flexible, human-like AI systems. Read More
Inside Standard Chartered’s approach to running AI under privacy rulesAI News For banks trying to put AI into real use, the hardest questions often come before any model is trained. Can the data be used at all? Where is it allowed to be stored? Who is responsible once the system goes live? At Standard Chartered, these privacy-driven questions now shape how AI systems are built, and
The post Inside Standard Chartered’s approach to running AI under privacy rules appeared first on AI News.
For banks trying to put AI into real use, the hardest questions often come before any model is trained. Can the data be used at all? Where is it allowed to be stored? Who is responsible once the system goes live? At Standard Chartered, these privacy-driven questions now shape how AI systems are built, and
The post Inside Standard Chartered’s approach to running AI under privacy rules appeared first on AI News. Read More
Gallup Workforce shows details of AI adoption in US workplacesAI News Artificial intelligence has moved into the US workplace, but its adoption remains uneven, fragmented, and tied to role, industry, and organisation. Findings from a Gallup Workforce survey covering the period to the end of December 2025 show how employees use AI, who benefits most from it, and where areas of uncertainty remain. The findings draw
The post Gallup Workforce shows details of AI adoption in US workplaces appeared first on AI News.
Artificial intelligence has moved into the US workplace, but its adoption remains uneven, fragmented, and tied to role, industry, and organisation. Findings from a Gallup Workforce survey covering the period to the end of December 2025 show how employees use AI, who benefits most from it, and where areas of uncertainty remain. The findings draw
The post Gallup Workforce shows details of AI adoption in US workplaces appeared first on AI News. Read More
Meta has started rolling out a new WhatsApp lockdown-style security feature designed to protect journalists, public figures, and other high-risk individuals from sophisticated threats, including spyware attacks. […] Read More
Google on Tuesday revealed that multiple threat actors, including nation-state adversaries and financially motivated groups, are exploiting a now-patched critical security flaw in RARLAB WinRAR to establish initial access and deploy a diverse array of payloads. “Discovered and patched in July 2025, government-backed threat actors linked to Russia and China as well as financially motivated Read […]
Cybersecurity researchers have discovered two malicious packages in the Python Package Index (PyPI) repository that masquerade as spellcheckers but contain functionality to deliver a remote access trojan (RAT). The packages, named spellcheckerpy and spellcheckpy, are no longer available for download, but not before they were collectively downloaded a little over 1,000 times. “Hidden inside the Read […]
When security teams discuss credential-related risk, the focus typically falls on threats such as phishing, malware, or ransomware. These attack methods continue to evolve and rightly command attention. However, one of the most persistent and underestimated risks to organizational security remains far more ordinary. Near-identical password reuse continues to slip past security controls, often Read More
Can We Trust LLM Detectors?cs.AI updates on arXiv.org arXiv:2601.15301v2 Announce Type: replace-cross
Abstract: The rapid adoption of LLMs has increased the need for reliable AI text detection, yet existing detectors often fail outside controlled benchmarks. We systematically evaluate 2 dominant paradigms (training-free and supervised) and show that both are brittle under distribution shift, unseen generators, and simple stylistic perturbations. To address these limitations, we propose a supervised contrastive learning (SCL) framework that learns discriminative style embeddings. Experiments show that while supervised detectors excel in-domain, they degrade sharply out-of-domain, and training-free methods remain highly sensitive to proxy choice. Overall, our results expose fundamental challenges in building domain-agnostic detectors. Our code is available at: https://github.com/HARSHITJAIS14/DetectAI
arXiv:2601.15301v2 Announce Type: replace-cross
Abstract: The rapid adoption of LLMs has increased the need for reliable AI text detection, yet existing detectors often fail outside controlled benchmarks. We systematically evaluate 2 dominant paradigms (training-free and supervised) and show that both are brittle under distribution shift, unseen generators, and simple stylistic perturbations. To address these limitations, we propose a supervised contrastive learning (SCL) framework that learns discriminative style embeddings. Experiments show that while supervised detectors excel in-domain, they degrade sharply out-of-domain, and training-free methods remain highly sensitive to proxy choice. Overall, our results expose fundamental challenges in building domain-agnostic detectors. Our code is available at: https://github.com/HARSHITJAIS14/DetectAI Read More
More at Stake: How Payoff and Language Shape LLM Agent Strategies in Cooperation Dilemmascs.AI updates on arXiv.org arXiv:2601.19082v1 Announce Type: new
Abstract: As LLMs increasingly act as autonomous agents in interactive and multi-agent settings, understanding their strategic behavior is critical for safety, coordination, and AI-driven social and economic systems. We investigate how payoff magnitude and linguistic context shape LLM strategies in repeated social dilemmas, using a payoff-scaled Prisoner’s Dilemma to isolate sensitivity to incentive strength. Across models and languages, we observe consistent behavioral patterns, including incentive-sensitive conditional strategies and cross-linguistic divergence. To interpret these dynamics, we train supervised classifiers on canonical repeated-game strategies and apply them to LLM decisions, revealing systematic, model- and language-dependent behavioral intentions, with linguistic framing sometimes matching or exceeding architectural effects. Our results provide a unified framework for auditing LLMs as strategic agents and highlight cooperation biases with direct implications for AI governance and multi-agent system design.
arXiv:2601.19082v1 Announce Type: new
Abstract: As LLMs increasingly act as autonomous agents in interactive and multi-agent settings, understanding their strategic behavior is critical for safety, coordination, and AI-driven social and economic systems. We investigate how payoff magnitude and linguistic context shape LLM strategies in repeated social dilemmas, using a payoff-scaled Prisoner’s Dilemma to isolate sensitivity to incentive strength. Across models and languages, we observe consistent behavioral patterns, including incentive-sensitive conditional strategies and cross-linguistic divergence. To interpret these dynamics, we train supervised classifiers on canonical repeated-game strategies and apply them to LLM decisions, revealing systematic, model- and language-dependent behavioral intentions, with linguistic framing sometimes matching or exceeding architectural effects. Our results provide a unified framework for auditing LLMs as strategic agents and highlight cooperation biases with direct implications for AI governance and multi-agent system design. Read More
AgenticSCR: An Autonomous Agentic Secure Code Review for Immature Vulnerabilities Detectioncs.AI updates on arXiv.org arXiv:2601.19138v1 Announce Type: cross
Abstract: Secure code review is critical at the pre-commit stage, where vulnerabilities must be caught early under tight latency and limited-context constraints. Existing SAST-based checks are noisy and often miss immature, context-dependent vulnerabilities, while standalone Large Language Models (LLMs) are constrained by context windows and lack explicit tool use. Agentic AI, which combine LLMs with autonomous decision-making, tool invocation, and code navigation, offer a promising alternative, but their effectiveness for pre-commit secure code review is not yet well understood. In this work, we introduce AgenticSCR, an agentic AI for secure code review for detecting immature vulnerabilities during the pre-commit stage, augmented by security-focused semantic memories. Using our own curated benchmark of immature vulnerabilities, tailored to the pre-commit secure code review, we empirically evaluate how accurate is our AgenticSCR for localizing, detecting, and explaining immature vulnerabilities. Our results show that AgenticSCR achieves at least 153% relatively higher percentage of correct code review comments than the static LLM-based baseline, and also substantially surpasses SAST tools. Moreover, AgenticSCR generates more correct comments in four out of five vulnerability types, consistently and significantly outperforming all other baselines. These findings highlight the importance of Agentic Secure Code Review, paving the way towards an emerging research area of immature vulnerability detection.
arXiv:2601.19138v1 Announce Type: cross
Abstract: Secure code review is critical at the pre-commit stage, where vulnerabilities must be caught early under tight latency and limited-context constraints. Existing SAST-based checks are noisy and often miss immature, context-dependent vulnerabilities, while standalone Large Language Models (LLMs) are constrained by context windows and lack explicit tool use. Agentic AI, which combine LLMs with autonomous decision-making, tool invocation, and code navigation, offer a promising alternative, but their effectiveness for pre-commit secure code review is not yet well understood. In this work, we introduce AgenticSCR, an agentic AI for secure code review for detecting immature vulnerabilities during the pre-commit stage, augmented by security-focused semantic memories. Using our own curated benchmark of immature vulnerabilities, tailored to the pre-commit secure code review, we empirically evaluate how accurate is our AgenticSCR for localizing, detecting, and explaining immature vulnerabilities. Our results show that AgenticSCR achieves at least 153% relatively higher percentage of correct code review comments than the static LLM-based baseline, and also substantially surpasses SAST tools. Moreover, AgenticSCR generates more correct comments in four out of five vulnerability types, consistently and significantly outperforming all other baselines. These findings highlight the importance of Agentic Secure Code Review, paving the way towards an emerging research area of immature vulnerability detection. Read More
