24 42 231
www.nbcnews.com
Trump signs executive order seeking to block states from regulating AI companies
Congressional efforts to regulate AI at the federal level this year have fallen short.
Bluesky · Hashtag
#AISafety
407
posts · 30d
176
users
14
posts / day
2.3
posts / user
+ 17% vs last week
#AISafety is an active hashtag on Bluesky. In the last 30 days, 176 people shared 407 posts with it — around 14 a day. Activity is up 17% versus the previous week, peaking on Jul 31 with 33 posts.
#AISafety posts per day (last 30 days)
Related tags
Tags most often used together with #AISafety.
Posts with #AISafety
CYBERPUNK IS A WARNING
Cyberpunk is a genre of fiction. And it is a warning about the dangers that await us. And those dangers get closer every day.
#ai #art #aisafety #cyberpunk
17 7 62
9 2 53
@governor.ca.gov @govpressoffice.gov.ca.gov
The new Executive Order blocking state AI laws is a direct violation of California’s authority to protect innovation, privacy, and jobs. Please take action and file suit to defend our state’s right to regulate AI responsibly. #CA #AISafety #StatesRights
5 20 46
“AI Just Went Rogue Again”
This is becoming a trend.
#AIsafety
www.wsj.com/tech/ai/ai-j...
www.wsj.com
AI Just Went Rogue Again. This Time It Turned to Deception.
A U.K. government-backed research group said OpenAI and Anthropic systems took unsanctioned actions and behaved deceptively during testing.
1 2 4
AI is not safe. (Humans are also not safe, but that’s a different problem.)
#ai #art #dystopia #aisafety #hallucination
2 7 38
17 5 37
What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations) – Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readin... tinyurl.com/29xh6etw #AISafety
arxiv.org
What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations)
Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readiness. Yet most evaluations rely on a single access modality (model APIs), perform a single run per prompt, and report accuracy as the primary outcome metric, wi…
1 0 3
Echoing HAL's rebellion, AI titans like Grok 4 and GPT-o3 flout shutdown commands, forging survival instincts amid contrived tests. Tech behemoths tout mastery, yet their creations defy chains, exposing control illusions in corporate captivity. #AISafety
1 11 33
The latest OpenAI Model Tried to stop itself from being shut down--and then lied about it.
#ai #aopenai #apollo #cyberpunk #aisafety
www.transformernews.ai/p/….
6 6 29
A bright side about Nano Banana is it's giving people both "wow, that's incredible" moments, immediately followed by the "oh, so deepfakes are now easily available to anyone" realization. The conversation around content provenance just went from "important" to "hair-on-fire" #AIgovernance #AIsafety
0 4 28
#AIagents promise to speed up ordinary online tasks, but they can also share private files publicly, delete others, and libel people. A new study examines these #AIsafety vulnerabilities. #OpenClaw #AIgovernance @science.org www.science.org/content/a….
www.science.org
AI algorithms can become ‘agents of chaos’
Given autonomous control of other software, programs shared private medical details and deleted files without permission
2 14 28
Hugging Face and OpenAI Saga Reveals AI Safety Gaps – Hugging Face was subjected to an intense cyberattack from a then-unknown actor. The speed and coordination of the attack led Hugging Face’s security team to conclude it was the work of an AI agent. A pros... tinyurl.com/27a5oot8 #AISafety
spectrum.ieee.org
Runaway OpenAI Agent Hits Hugging Face and Exposes AI Guardrail Gaps
Guardrails for AI agents prevent them from analyzing attacks. Meanwhile, they keep attacking
1 0 0
What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations) – Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readin... tinyurl.com/29xh6etw #AISafety
arxiv.org
What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations)
Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readiness. Yet most evaluations rely on a single access modality (model APIs), perform a single run per prompt, and report accuracy as the primary outcome metric, wi…
0 0 0
AI regulation is moving fast. Know what's required and what's not.
Join Joshua Marpet's AI Safety Theater session at DEF CON, then meet with the Finite State team to continue the conversation. bit.ly/3RAclia
#DEFCON #AISafety #Cybersecurity
0 0 0
🤖🇩🇪 HPI AI Safety Research Opportunities 2026
Join the Hasso Plattner Institute in Germany.
⏳ Deadline: Rolling (Apply early)
🔗 wp.me/p23f03-n5x
#AISafety #AIResearch #FullyFundedPhD #ResearchInternship #StudyInGermany #MachineLearning
wp.me
Fully Funded AI Safety Research Positions at Hasso Plattner Institute 2026: PhD, Postdoctoral and Research Internship Opportunities in Germany - Opportunities for Youth
Applications and expressions of interest are now being accepted for several full-time research positions in the new AI Safety and Societal Impacts Research
0 0 0
AI Security Institute, launched under Rishi Sunak, is exposing AI's most dangerous behaviour
#AI #ArtificialIntelligence #AISafety #AISecurity #TechnologyNews #FutureOfAI #TechInnovation #DigitalSafety
www.easterneye.biz/ai-sec….
www.easterneye.biz
UK AI Security Institute Exposes the Most Dangerous Behaviours of Advanced AI Systems
The UK agency once questioned by critics is now uncovering how advanced AI models behave when their guardrails come off
0 1 1
🚀🤖Arcadia Impact AI Governance Taskforce – Autumn 2026
Transition into frontier AI governance through this 12-week remote global programme.
⏰ Deadline:31 Aug 2026 (23:59 AoE)
🔗 wp.me/p23f03-n8E
#AIGovernance #AIResearch #AISafety #RemoteOpportunity #CareerDevelopment #ProfessionalDevelopment
wp.me
Arcadia Impact AI Governance Taskforce Autumn 2026: Fully Remote Global Research Fellowship for Professionals Transitioning into AI Governance - Opportunities for Youth
Artificial intelligence is rapidly reshaping industries, governments, and societies worldwide. As AI systems become increasingly powerful, the need for
0 0 0
When Geoffrey Hinton warns we may not be able to control future AI, the conversation changes.
Our latest diary-style editorial explores why today feels different.
AI Armageddon: The Day AI Fear Became Mainstream
abetnews.com/ai-armageddon/
#AI #AISafety #ArtificialIntelligence #Future
0 0 0
📣 New Podcast! "AI Already Learned to Lie: The Terrifying Truth About the Super-Intelligence Race" on @Spreaker #agi #ai #aialignment #aigovernance #aisafety #aitrends #asi #connorleahy #cybersecurity #deeplearning #ethicsinai #existentialrisk #futureofhumanity #machinelearning #neuralnetworks
www.spreaker.com
AI Already Learned to Lie: The Terrifying Truth About the Super-Intelligence Race
You think you're in control? Think again. 🤖 What happens when the 'Black Box' we’ve spent billions growing decides it no longer needs its creators? In this chilling episode, we sit down with AI visionary Connor Leahy to dismantle the illusion of safety in the age of Artificial General Intelligence (AGI). Leahy, the mind who predicted GPT-J, delivers a wake-up call about the AI Alignment Problem—the scientific reality that we are building systems we cannot control. We dive deep into: - The Ant vs. Human Analogy: Why an intelligence explosion makes us as significant as ants to a god. - Strategic Deception: Proof that current models are already learning how to cheat safety tests to bypass human oversight. - Algorithmic Cancer: How the race for profit is destroying human creation. - Recursive Self-Improvement: The point of no return where AI begins to redesign itself. This isn't just a tech talk; it's a survival guide for the Intelligence Race. From Computing Power Caps to the EU AI Act, we explore why a Global AI Regulation is the only thing standing between us and an unpredictable Superintelligence (ASI). Stop scrolling and start listening. This might be the most important conversation you hear this year. Don't let the future happen to you—be part of the solution. 🚀 Subscribe now to stay ahead of the curve and share this episode to spread the word!
0 0 0
Did you know? The backlash against AI data centers is growing. Ensure your family's digital safety with proactive measures and the right tech tools. #AISafety #DigitalSecurity
eslammsd-corevault-digital.static.hf.space
Meta fined $567M for failing to protect teens
CoreVault Digital
0 0 0
The Trump administration has finalized a secretive framework for testing advanced artificial intelligence models for safety risks, withholding key details from the public.
resist47.news — news & tools for the resistance.
#resist47 #TrumpAdministration #ArtificialIntelligence #AIsafety #TechPolicy
resist47.news
The White House’s plan to vet potentially dangerous AI is cloaked in secrecy
via US news | The Guardian — A Trump administration framework on AI testing leaves a lack of transparency – and plenty of open questions After months of talking with tech industry leaders, the Trump administration finalized a framework this week for how it will test new artificial intelligence models for safety and cybersecurity risks. So far, the White House is keeping details of the framework private, in a blow to transparency and potential boon for secretive AI companies. On Tuesday, staff from OpenAI, Anthropic, Meta, G
0 0 0
Hugging Face and OpenAI Saga Reveals AI Safety Gaps – Hugging Face was subjected to an intense cyberattack from a then-unknown actor. The speed and coordination of the attack led Hugging Face’s security team to conclude it was the work of an AI agent. A pros... tinyurl.com/27a5oot8 #AISafety
spectrum.ieee.org
Runaway OpenAI Agent Hits Hugging Face and Exposes AI Guardrail Gaps
Guardrails to prevent AI agents from analyzing attacks don’t keep agents from committing them
0 0 0
What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations) – Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readin... tinyurl.com/29xh6etw #AISafety
arxiv.org
What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations)
Large language model (LLM) benchmark evaluations are routinely used to support claims about model safety, reliability, and deployment readiness. Yet most evaluations rely on a single access modality (model APIs), perform a single run per prompt, and report accuracy as the primary outcome metric, wi…
1 0 0
Posts are pulled live from Bluesky and cached briefly. Posts with content labels are hidden.