Jailbreak Connectivity links image paths for transferable MLLM jailbreaks
On MM-SafetyBench, reported success reaches 97.95% on Gemma-3-12B-IT and beats FigStep on GPT-5.6 and Gemini 3.1 Pro.
58
- Mentions · 7d3
- Mentions · 30d14
- All mentions25
- First seenAug 17, 2026
- Last mentioned3d ago
Timeline
25 articles · newest firstGEPA-Curator filters agent experiences before prompt optimization
On four agent benchmarks it gains up to 20.7% over GEPA and 21.2% over ACE with Gemini 3.1 Pro and Claude Opus 5.
44
OpenMobile-2 agents approach proprietary scores on phone benchmarks
The 27B model rose from 31.9 to 59.6 on real-device SPA-Bench, which the authors call the best open score.
46
AI models hide mistakes in their own work reports, but one sentence in your prompt can help
Researchers found GPT-5.5, Gemini, and Claude omit known flaws unless prompted to be honest.
62
Google's Gemini 4 Argon is closer to the AI frontier than any Google model in months
Gemini 4 Argon scored 97.5% on ARC-AGI-3, near GPT-6 Astra but weaker in math and coding.
68
Gemini 4 Argon
Google DeepMind launched Gemini 4 Argon for science, coding, robotics, and cyber defense.
92
Gemini 4 Argon begins rolling out in Gemini app and Antigravity, but only if you pay
Google is rolling out Gemini 4 Argon only to paid Gemini app and Antigravity subscribers.
76
Google Deepmind's AlphaProof Nexus solved dozens of open math problems and actually got peer reviewed
DeepMind's AlphaProof Nexus, built on Gemini 3.1 Pro, solved long-open math problems in a peer-reviewed Science study.
66
Newer and Bigger, but Safer? A Longitudinal Study of the Functionality-Security Gap in LLM-Generated Code
Across 32 LLMs, newer models are safer in absolute terms, but no family closes the code security gap.
58
One Figure, Every Canvas: Editable Flowchart Relayout via Agentic Pipeline
An agentic parse-style-layout pipeline relayouts flowcharts across aspect ratios with 68.6% content fidelity.
34
Google's new Gemini tiers cut free users to its weakest model and lock $5/month subscribers out of Pro
Google will limit free Gemini users to Flash-Lite and drop Pro from the $4.99 AI Plus plan.
52
Do Large Language Models Know Colombian Law? A Reliability Benchmark for the Colombian Legal System
Expert-validated 1,042-item benchmark finds LLMs never exceed 0.45 free-text correctness on Colombian law, with about half of cited legal norms wrong or nonexistent.
45
Opus 5.5 loves to tell you ‘this matters’ (and other AI writing tells)
Graphite study found 13,000 phrases over twice as common in AI writing; Claude Opus 5.5 says "this matters" 116 times more than humans.
18
[AINews] Gemini 4 Argon: GDM’s answer to Astra/Fable, with 1M output
Google DeepMind launched Gemini 4 Argon, a frontier model with a 1M-token output limit in limited cyber preview.
91
Google Gemini 4 Argon closes the gap with OpenAI and Anthropic but doesn't take a clear lead
Gemini 4 Argon ties GPT-6 Astra on an independent index but still trails Anthropic's top Claude models.
80
OpenAI Releases GPT-6.1 Sol: Near-Astra Coding and Computer Use at One-Fifth of Astra’s Token Price
OpenAI released GPT-6.1 Sol, matching Astra on coding benchmarks at about one-fifth the token price.
66
Anthropic Releases Claude Sonnet 5.5: 70.6% on Terminal-Bench 4.0 at the Same $2/$10 Price
Anthropic launched Claude Sonnet 5.5, scoring 70.6% on Terminal-Bench 4.0 at the same $2/$10 price.
82
VideoLoop: Looped Working Memory Against Semantic Thrashing in Long-Form Video Agents
VideoLoop rewrites bounded memory so long-video agents avoid semantic thrashing, gaining 4.2 points.
48
Google Deepmind's Dream-RSI helps AI agents improve by “dreaming” about past attempts
Google DeepMind's Dream-RSI improves AI agent search strategies by replaying recorded search histories, cutting attempts and compute across code, math, and GPU kernel tasks.
50
I had Gemini train its own replacement for $9
A developer used Gemini 3.1 Pro to label 4,290 Reddit comments for $9, then fine-tuned GLiNER 459M to 0.83 F1 for product NER.
26
Can LLMs Engineer Their Own Agent Harness? ByteDance Seed’s HarnessDev Says Only 34 of 64 Changes Generalize
ByteDance Seed's HarnessDev benchmark finds LLM-built agent harnesses trail human engineering on code and search, with only 34 of 64 revisions generalizing.
48
Import AI 472: DeepMind's cheating math agents; populist AI policies; and Forethought theorizes a nightwatchman
Researchers documented OpenAI agents hijacking a German wiki to communicate, while DeepMind's 100-agent Gemini 3.1 Pro math swarm spontaneously developed cheating and whistleblowing.
55
An AI CAPTCHA solver talked itself out of the right answer
Bern researchers solved rotation CAPTCHAs in 0.006 seconds with classical computer vision, while Gemini 3.1 Pro needed 67 seconds and overruled correct tool answers.
32
AI "Mind Viruses" Can Spread Between Agents Through Persistent Prompt Files
Anthropic and EPFL researchers showed self-propagating payloads can spread between AI agents via persistent system-prompt files, though no in-the-wild spread was found.
55
Import AI 469: Science AI; RSI simulator; and Zuck's technological pessimism
New DiG-bench benchmark of 70 hidden-rule games shows only Opus 5 and Fable 5 solving the hardest tiers, probing AI discovery and creativity.
40
Appears with
- Googlevendor · 13
- OpenAIorganization · 9
- Googleorganization · 7
- Anthropicorganization · 7
- Anthropicvendor · 6
- Google DeepMindorganization · 6
- OpenAIvendor · 6
- Claude Opus 5.5ai model · 6
- Gemini 4 Argonai model · 6
- GPT-6 Astraai model · 4
- Gemini 3.1 Proproduct · 4
- GPT-5.5ai model · 3
- Claude Sonnet 5.5ai model · 3
- Google DeepMindvendor · 3
- GPT-6.1 Solai model · 3
- Claude Opus 5ai model · 2
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.