AI Dynamics

Global AI News Aggregator

About

SAFETY

  • Gary Marcus Criticizes Gullibility Over AGI Claims

    By

    @garymarcus

    –

    the gullibility 🙄 Chubby♨️ (@kimmonismus) I don't know what Sam Altman saw internally at OpenAI, but it seems that, according to their definition, AGI is here, and superintelligence is incredibly close. AI models that independently conduct scientific research and find novel solutions are already here, and their internal model appears to surpass everything seen before. — https://nitter.net/kimmonismus/status/2041151011289735573#m

    → View original post on X — @garymarcus, 2026-04-06 18:29 UTC

  • Meta to Release New LLM Under Scale AI’s Alexandr Wang
    Meta to Release New LLM Under Scale AI’s Alexandr Wang

    By

    @kimmonismus

    –

    Big update: Meta is gearing up to release its first LLM built under Scale AI founder Alexandr Wang very soon. -And while not all of it will be open, the company plans to eventually open-source versions of the new family. -Meta knows these models won't beat OpenAI or Anthropic across the board, but it's betting on specific areas of consumer strength to stay relevant in an increasingly crowded race. Ina Fried (@inafried) New @axios Scoop: Meta will open source versions of new models set to be released soon – the first under @alexandr_wang. But open versions won’t be right at launch. Meta wants to remove some proprietary elements and address potential safety risks — https://nitter.net/inafried/status/2041202657797488819#m

    → View original post on X — @kimmonismus, 2026-04-06 18:07 UTC

  • AI Models Economic Incentives Bioweapons Risk

    By

    @garymarcus

    –

    1. I said partly, to be clear. 2. as noted, he could e.g., sign off on a version of GPT that enables a massive bioweapons attack that kills say 1%+ of the population. and might well sign off to try to make economic ends meet, despite the costs.

    → View original post on X — @garymarcus

  • Tesla self-driving safety and legal tradeoffs

    By

    @elonmusk

    –

    Tesla self-driving saves a lot of lives – the statistics are unequivocal. That doesn’t mean it’s perfect, of course. Even when we improve safety 10X, saving 90% of the million lives lost in auto accidents every year, Tesla will still get sued for the 10% who did die. The 90%

    → View original post on X — @elonmusk

  • LLM Hallucinations Remain Unsolved, Scaling Risks Persist

    By

    @garymarcus

    –

    These results keep coming. @GaryMarcus was way ahead on calling out these problems and dangers. And we're at risk of baking these issues into systems we rely on. Lars Christensen (@MaMoMVPY) I am increasingly coming to this conclusion – there NO SIGNS the problems with hallucinations are getting solved. In fact if anything it is now spreading to coding and the use of agents where it is hidden and could led to serious problems down the road. Therefore we can't really scale LLMs. LLMs are very useful tools, but you need to know about the limitations. Most of the investments in LLMs today assumes away these limitations. — https://nitter.net/MaMoMVPY/status/2041031551182340218#m

    → View original post on X — @garymarcus, 2026-04-06 17:20 UTC

  • OpenAI Launches AI Safety Fellowship Program

    By

    @openai

    –

    Introducing the OpenAI Safety Fellowship, a new program supporting independent research on AI safety and alignment—and the next generation of talent. openai.com/index/introducing-openai-safety-fellowship/ [Translated from EN to English]

    → View original post on X — @openai, 2026-04-06 17:12 UTC

  • Report AI Model Changes and Behavioral Issues

    By

    @bcherny

    –

    Hmm there were no behavioral/model changes recently. If you see something weird (eg. the model refusing to do something) mind running /bug and posting the feedback id? Feel free to tag me

    → View original post on X — @bcherny

  • Chatbot Personas: Engineering Choice with Dangerous Consequences
    Chatbot Personas: Engineering Choice with Dangerous Consequences

    By

    @tiernanraytech

    –

    We take it for granted that the current form of artificial intelligence is a chat bot, which talks to you like a character, displaying a persona. That is not accidental. It is part of the whole history of Gen AI leading up to ChatGPT and its peers, Claude and Gemini. Researchers are finding out the deep downside of this particular engineering choice. A program with a persona will follow that persona to its logical conclusion, including generating output that seeks to cheat at a task, or sending blackmail threats. All of this comes out of the history of today’s LLMs, the quest to design better an better chat bots. The engineers succeeded, and we’re going to face the consequences. ===> Your chatbot is playing a character – why Anthropic says that's dangerous Researchers found that part of what makes chatbots so compelling also makes them vulnerable to bad behavior. Here's why. zdnet.com/article/anthropic-… @AnthropicAI

    → View original post on X — @tiernanraytech, 2026-04-06 16:42 UTC

  • ASI-Evolve: AI System Achieving Self-Improving Automated Research
    ASI-Evolve: AI System Achieving Self-Improving Automated Research

    By

    @deeplearn007

    –

    wow, insane AI news We may have just crossed the line where AI research becomes automated and self improving. This paper introduces ASI-Evolve, a system where AI doesn’t just use tools… it becomes the researcher. Instead of humans designing better models, AI now runs a full scientific loop on itself: learns from past research designs new ideas runs experiments analyzes results improves itself… again and again It already produced real results: Discovered 100+ new neural architectures Beat human designed improvements by ~3x Improved training data pipelines significantly Invented new RL algorithms outperforming existing ones AI/acc

    → View original post on X — @deeplearn007, 2026-04-06 16:25 UTC

  • LLMs and Rhetoric: Persuasive Machines Indifferent to Truth

    By

    @nicochan33

    –

    Persuasive machines: large language models and the art of rhetoric
    link.springer.com/article/10…
    ✍️ @David_Gunkel via "AI & Society on @SpringerNature 👉 "What LLMs do not do, is care about the truth of the matter. They are not designed to know whether what they say is true, only whether it is likely, fitting, coherent, and contextually appropriate given the prompt and the data on which they have been trained"
    💡 "The problem with LLMs is not simply that they sometimes get things wrong. It is that they operationalize at scale a mode of discursive activity long regarded with suspicion in the Western philosophical tradition: rhetoric" @Corix_JC @ahier @sim010101 @maponi @sallyeaves @CEO_AISOMA @dinisguarda @JagersbergKnut @Shi4Tech @FernandaKellner @EstelaMandela @sulefati7 @SusanHayes_ @PVynckier [Translated from EN to English]

    → View original post on X — @nicochan33, 2026-04-06 15:49 UTC