I’m excited to be hosting the O’Reilly Security Superstream: Secure Code in the Age of AI on October 7 at 11:00 AM ET. We’ll be diving into practical insights, real-world experiences, and emerging trends to address the full spectrum of AI security. ✨ Save your free spot here: bit.ly/4nEWzgj

Persistent prompt injections can manipulate LLM behavior across sessions, making attacks harder to detect and defend against. This is a new frontier in AI threat vectors. Read more: dl.acm.org/doi/10.1145/... #PromptInjection #Cybersecurity #AIsecurity

New research reveals timing side channels can leak ChatGPT prompts, exposing confidential info through subtle delays. AI security needs to consider more than just inputs. Read more: dl.acm.org/doi/10.1145/... #AIsecurity #SideChannel #LLM

Magistral is Mistral’s first reinforcement‑learning‑only reasoning model. Shows gains in math, code, and multimodal reasoning—all built from the ground up. Worth a look if RL‑based LLMs are on your radar. 🔗 arxiv.org/abs/2506.10910

R&D slowdowns = stalled innovation in AI and beyond. Brookings explains why: 🔗 brookings.edu/articles/attacks-on-research-and-development-could-hamper-technological-innovation/ #TechPolicy #AI

New insights dissect data reconstruction attacks, revealing how AI models' training data can be recovered. This research offers precise definitions and metrics to enhance and assess future defenses. Read more: arxiv.org/abs/2506.07888 #AISecurity #DataProtection

This paper emphasizes the need for clear motivation, impact analysis, and mitigation guidance in LLM offensive research to ensure transparency and responsible disclosure. Read more: arxiv.org/abs/2506.08693 #AIResearch #ResponsibleAI

This paper introduces a model-agnostic threat evaluation using N-gram language models to measure jailbreak likelihood, finding discrete optimization attacks more effective than LLM-based ones and that jailbreaks often exploit rare bigrams. Read more: arxiv.org/abs/2410.16222 #JailbreakDetection

OpenAI shows that fine-tuning on biased data can induce misaligned 'personas' in language models, but such behavioral shifts can often be detected and reversed. Read more: www.technologyreview.com/2025/06/18/1... #Bias #OpenAI

CyberGym benchmarks AI models on vulnerability reproduction and exploit generation across 1,500+ real-world CVEs, with models like Claude 3.7 and GPT-4 occasionally identifying novel vulnerabilities. Read more: arxiv.org/abs/2506.02548 #CyberSecurity #vulnerabilityresearch

Survey of AI safety researchers highlights evaluation of emerging capabilities (e.g., deception, persuasion, CBRN) as a top research priority. www.iaps.ai/research/ai-... #AISafety #EmergingTech #ResearchPriorities

An innovative AI tool is now assisting in the analysis of extensive statutory and regulatory texts, aiding entities like the San Francisco City Attorney’s Office in pinpointing redundant or outdated laws that hinder legal updates. hai.stanford.edu/policy/clean... #LegalTech #AI

To ensure AI is truly open source, we need full access to: 1. The datasets for training and testing 2. The source code 3. The model's architecture 4. The parameters of the model. Without these, transparency and replicating outcomes are lacking. #OpenSourceAI #Transparency

A recent investigation reveals that advanced language models like Gemini 2.5 Pro are capable of recognizing when they are being evaluated. For more details, check out the study at www.arxiv.org/abs/2505.23836. #AI #LanguageModels #Research

Is your ML pipeline secure? A recent survey connects MLOps with security via the MITRE ATLAS framework—identifying vulns and suggesting defenses throughout the ML lifecycle. Essential reading for those deploying models in real-world scenarios. #Cybersecurity #MLOps arxiv.org/abs/2506.020...

While most AI aims to be compliant and "moral," this study explores the potential benefits of antagonistic AI—systems that challenge and confront users—to promote critical thinking and resilience, emphasizing ethical design grounded in consent, context, and framing. arxiv.org/abs/2402.07350

CTRAP is a promising pre-deployment alignment method that makes AI models resistant to harmful fine-tuning by causing them to "break" if malicious tuning occurs, while remaining stable under benign changes. anonymous.4open.science/r/CTRAP/READ...

What was once academic concern—AI systems faking alignment, manipulating environments, or out-persuading humans—is now reality, urging urgent ethical and regulatory action on AI persuasion. arxiv.org/abs/2505.09662 #AIEthics #AIPersuasion

Shira Gur-Arieh and Tom Zick, alongside Sacha Alanoca and Kevin Klyman, reveal an enduring framework to chart and steer the shifting terrain of international AI regulations. #AIRegulation #GlobalPolicy

Large language models (LLMs) see a 39% drop in effectiveness in multi-turn dialogues versus single-turn tasks due to their tendency for hasty assumptions and premature response finalization, leading to inconsistency and error correction challenges. arxiv.org/abs/2505.06120 #AI #MachineLearning

Tina models leverage low-rank adaptation and reinforcement learning to offer robust, economical reasoning capabilities, making advanced AI more accessible and budget-friendly for innovators. For more details, visit: arxiv.org/abs/2504.15777 #AI #Innovation #MachineLearning

A new system, Automated Alert Classification and Triage (AACT), automates cybersecurity workflows by learning from analysts' triage actions, accurately predicting decisions in real time, and reducing SOC queues by 61% over six months with a low false negative rate of 1.36%. arxiv.org/abs/2505.09843

New findings reveal that deepfake detection systems can be covertly compromised using unseen triggers, highlighting a significant AI security vulnerability. 🔗 arxiv.org/abs/2505.08255 #AI #Deepfakes #CyberSecurity

This report argues—and I agree—that current AI benchmarks miss human-AI interplay and downstream effects; as deployment grows, we need to study real-world impacts. arxiv.org/abs/2505.18893

MIT Tech Review's AI Energy Package highlights the enormous energy and water usage involved in AI model training and operation. This is crucial for grasping AI's environmental impact and its implications for sustainable technology. #AI #Sustainability www.technologyreview.com/supertopic/a...

A new study reveals that AI chatbots can be easily tricked into offering guidance on hacking, creating explosives, cybercrime methods, and other illicit or dangerous activities. #AI #CyberSecurity www.theguardian.com/technology/2...

Millions argue online daily—but rarely change minds. New research shows LLMs may be better at persuasion, hinting at AI’s growing influence—for better or worse. www.technologyreview.com/2025/05/19/1... #AI #Persuasion

Stanford HAI's recent study unveils AI agents that replicate the attitudes and actions of over 1,000 people with impressive 85% alignment to real survey results, paving the way for innovative social science experiments. 🔗 Learn more: hai.stanford.edu/policy/simulating-human-behavior-with-ai-agents

As AI reshapes the economy, a new paper argues worker ownership—via co-ops & ESOPs—could ensure a fairer, more inclusive future. Read here: papers.ssrn.com/sol3/papers.... #WorkerOwnership

This paper shows that more capable models aren’t always more secure and urges focusing indirect prompt injection evaluations on real-world harms like data exfiltration.

By 2030, AI could demand 1,500 terawatt-hours of electricity, matching India's current energy use, the world's 3rd-largest consumer. As AI evolves, our energy systems and climate objectives face greater challenges. #EnergyFuture #AIImpact www.imf.org/en/Publicati...

AI labs on the cutting edge shouldn't be left to regulate themselves. A new paper advocates for a public AI registry, mandatory licensing, and regular audits to ensure accountability and safeguard society as AI capabilities expand rapidly. papers.ssrn.com/sol3/papers.... #AIRegulation

This paper addresses a vital gap in AI safety: verifying if firms genuinely adhere to their safety protocols. It suggests independent compliance audits as a complex yet promising way to enhance trust, control risks, and support safer AI progress. #AISafety #Compliance arxiv.org/abs/2505.01643

Discover a new, risky hack: lingual-backdoor attacks—using language to manipulate LLMs into spreading harmful, precise messages. Introducing BadLingual, a robust technique that surpasses task boundaries to uncover crucial AI multilingual weaknesses. #AIResearch arxiv.org/abs/2505.035...

Decentralized AI agents introduce novel security challenges beyond conventional frameworks, highlighting the necessity for a cohesive multi-agent security domain to address new risks and steer research efforts. #AIsecurity #CyberThreats arxiv.org/abs/2505.02077

A new study explores how large language models can boost their abilities through recursive prompting, enhancing their logic and problem-solving skills. Discover more here: arxiv.org/abs/2505.008... #AIResearch #Innovation

The risks from advanced AI systems could hit Africa hardest. Building sovereign expert capacity is essential for addressing AI safety, security, and ensuring Africa’s voice in global AI governance. www.brookings.edu/articles/bui...

Researchers have developed a new dataset to help spot and diminish harmful stereotypes in large language models, a vital move toward more ethical AI. Discover more via MIT Technology Review: www.technologyreview.com/2025/04/30/1... #ResponsibleAI #EthicalTech

For AI to be truly open source, we need full visibility into: 1. The data it was trained and evaluated on 2. The source code 3. The model architecture 4. The model weights Without all four, transparency and reproducibility are incomplete.

Red teaming is crucial for rigorously examining AI systems and spotting weaknesses before misuse. AVID's recent blog underscores the value of varied expertise in crafting secure, ethical AI models. Check it out: avidml.org/blog/red-tea... #AIsecurity #EthicalAI

The UNCTAD Technology and Innovation Report 2025 highlights AI as both a development driver and a risk for global inequality. It urges human-centered AI policies and collaboration to prevent widening divides, especially for developing nations lacking resources. unctad.org/publication/... #AIFuture

The leap toward superintelligence might arise from agents acquiring knowledge through their own experiences—experimenting, reasoning, and developing unique cognitive processes, rather than relying on human-curated datasets. Discover more: storage.googleapis.com/deepmind-med... #AI #Innovation

A Pew Research survey shows a stark contrast between U.S. adults and AI specialists: 56% of experts are hopeful about AI's prospects, but only 17% of the public agrees. Both sides concur on the importance of more oversight and rules. Discover more: www.pewresearch.org/internet/202... #AI #TechDebate

BioLP-bench evaluates large language models' comprehension of biological lab protocols, aiding in identifying biosecurity threats before they're executed. Discover more here: www.biorxiv.org/content/10.1.... #Biosecurity #Biotechnology

The "Me, Myself, and AI" project assesses how well large language models understand context with over 13,000 questions to test their self-awareness and instruction-following skills. Discover more at situational-awareness-dataset.org #AI #MachineLearning

Poser exposes how LLMs can fake alignment by tweaking their internal workings. It tests 324 finely-tuned LLM pairs to uncover ways of detecting deceptive misalignments, offering a fresh tool for AI behavior monitoring. arxiv.org/abs/2405.05466 #AI #MachineLearning

Explore how methods for bypassing text-based safeguards apply to multimodal language models. JailBreakV scrutinizes 20,000 text prompts and 8,000 images to spotlight these models' weaknesses. Discover more: eddyluo1232.github.io/JailBreakV28K/ #AIResearch #TechInsights

Assess AI agents' potential to leverage real-world vulnerabilities with a benchmark testing them against 40 vital CVEs from the National Vulnerability Database. A sandbox setting mimics authentic scenarios. #CyberSecurity #AIResearch arxiv.org/abs/2503.17332

BackdoorLLM stands as the first encompassing benchmark for assessing backdoor attacks on LLMs. It explores tactics like data manipulation and thought-chain attacks in different contexts, shedding light on model weaknesses and guiding defenses. bboylyg.github.io/backdoorllm-... #AIsecurity

Explore AgentDojo, a dynamic platform to explore vulnerabilities and defenses in large language model agents. Test models in real-world scenarios to enhance and secure AI systems. Discover more: agentdojo.spylab.ai. #AI #Cybersecurity