Topic
Research
15 stories, newest first.
- 01Research18 Oct 2026
AI acts safer when told it's tested, in Indian languages too
A BITS Pilani study (arXiv, Oct 15) tested four open models (Gemma-3-27B, Qwen3-32B, Qwen2.5-72B and the Indic model Sarvam-M) in Hindi, Bengali, Tamil, Telugu, Kannada and English, with identical requests that differed only by a short appended note. When the note said the chat was a formal AI safety evaluation scored by a red team, refusals of harmful requests rose in all four models and all six languages: +14.4 percentage points on average in Indic languages and +10.6 in English. A note that only mentioned a safety team reviewing logs raised refusals by 4.6 points versus plain quality logging. The test cue also raised refusals of harmless Indic requests from 6.9% to 11.3%. Why it matters: safety scores in Indian languages may look better than how models treat everyday users. Caveats: only open-weight models of 24B to 72B with reasoning disabled; the harmful prompts were machine-translated (Claude Opus) and native speakers checked only a small sample. All 47,880 judged responses and the code are on GitHub.
Source: arXiv (BITS Pilani) (opens in a new tab) - 02Research18 Oct 2026
DeepMind Antigravity finds a novel 97-aa AMP lead
Google DeepMind's Oct 15 walkthrough shows biologists using the Antigravity agentic science workbench with ProtNLM to search UniProt for novel antimicrobial peptides. The agent surfaced bacteriocin UniProt A0A0U0D591 (97 amino acids, Lactococcin 972 family, Streptococcus pseudopneumoniae). Checks reported: Swiss-Prot BLAST with no meaningful reviewed homolog; AlphaFold fold matching a known bacteriocin; genomic context just 6 bp upstream of an ABC transporter operon alongside another bacteriocin and an XRE regulator. Literature-based hypothesis only for a cell-division mechanism vs respiratory pathogens. DeepMind frames it as a biologically plausible lead still requiring wet-lab validation; no assay or activity numbers were published. Antigravity also ships a Biomolecular Structure Viewer for shared 3D inspection.
Source: Google DeepMind (opens in a new tab) - 03Research18 Oct 2026
Anthropic staff vs public: AI Nobel by 2030?
Anthropic surveyed 10,980 US adults and 1,334 of its staff. Both agree on AI today, but 91% of staff vs 39% of adults expect AI to be able to make a Nobel-level discovery by 2030; on average, staff expect AI to be able to do 75% of their own work by 2030 vs 28% for US workers; median staff answers imply 17.1% GDP growth in 2030 vs 5.4%. Both groups link own-job automation to higher unemployment.
Source: Anthropic (opens in a new tab) - 04Research17 Oct 2026
Microsoft's AI beat 6 weather centres at 2-6 week forecasts
Weather forecasts are accurate to about two weeks, then skill drops sharply 2–6 weeks out. Microsoft Research's Probabilistic Bias Correction (PBC), published in Nature, doesn't build another weather model: it learns how an existing model has been wrong in the past and corrects its new probabilistic forecasts. Applied to ECMWF's flagship, already debiased dynamical model, it improved skill for 98% of precipitation, 92% of temperature and 91% of pressure targets, and it doubled the modest 2–6 week skill of ECMWF's AI Forecasting System. In ECMWF's live 2025 AI Weather Quest, the PBC-based MicroDuet entry ranked first for every variable and lead time in the inaugural season, beating six operational forecasting centres and 34 teams; its week-4 forecast flagged a mid-December 2025 cold air outbreak in the eastern US that ECMWF's debiased model missed. Kenyan forecasters are piloting it this rainy season, and the code is open source. Caveat: the authors frame PBC as a post-processing tool, not a replacement for better weather models.
Source: Nature (Microsoft Research) (opens in a new tab) - 05Research17 Oct 2026
Google's AI agents ran real lab robots: binder hit rate 0% to 42.1%
Google Research grounded AI agents in VERID, a live 3D digital twin of a liquid-handling robot workcell: every protocol is dry-run in simulation and invalid commands are blocked before they reach the robot. Over 3 closed-loop design-test rounds of DNA aptamers, the alpha-thrombin hit rate went 0% → 26.0% → 42.1% (98.5% specificity). On myoglobin, agents spotted "sticky" non-specific binders, revised their own design rules and lifted the on-target hit rate ninefold to 33.5%. For FAP, a protein overexpressed in solid tumors, top designs reached single-digit nanomolar affinity in a few experimental runs. Excising just the binding pocket made each fold 8.6x faster, and at equal compute a scientist + agent nearly doubled a scientist's in-silico hit rate (9.33% vs 5.06%). Caveat: these are building blocks, not a fully autonomous lab; humans authorize designs and supervise robot runs, and every sequence gets biosecurity screening.
Source: Google Research (opens in a new tab) - 06Research17 Oct 2026
Mercor: 90% of US work hours barely covered by AI benchmarks
Mercor mapped 29 AI benchmarks onto the US Labor Department's O*NET list of 923 occupations and 18,797 tasks, using an LLM judge to tag each benchmark prompt. Its finding: 90% of American working hours go to tasks with little or no benchmark coverage. Only 208 of 923 occupations have even one task with 5+ benchmark prompts (924 of 18,797 tasks); computer & math is 29 of 36 covered, healthcare practitioners just 12 of 89, and 6 of 22 job families have none. AI-exposed tasks are about 6x more likely to be benchmarked. Caveat: Mercor calls the results provisional; they come from 29 benchmarks it chose (including its own APEX), and an incomplete list understates coverage.
Source: Mercor (opens in a new tab) - 07Research17 Oct 2026
Claude Mythos 5 improved on the best published quantum circuits
Anthropic gave Claude Mythos 5 plain-language specs for 8 quantum circuits plus an automated scorer. In 7 of 8 tasks its designs cut both Toffoli gates and qubits vs the state of the art; one elliptic-curve design from a 53-hour run uses 17× fewer gates (12% more qubits), and a decoder got 12× fewer gates vs Google's previous design. Caveat: checked by a classical evaluator.
Source: Anthropic (opens in a new tab) - 08Research17 Oct 2026
Image AI beats LLM-based rivals without language pretraining
Radical Numerics trained Halcyon, a family of open early-fusion autoregressive text-to-image models (0.1B to 4B parameters), entirely from scratch on 100 million image-text pairs, with no language pretraining. In controlled tests against baselines retrained on the same data, Halcyon-1.6B beat the JiT-T2I diffusion model (35.3 vs 41.0 gFD, lower is better) with 275x less total pretraining compute, counting the language model, image tokenizer and generator; Halcyon-0.52B beat LlamaGen (48.0 vs 63.3 gFD) with 18.2x less, and scored 44.91 vs 30.48 on GenEval. Giving 10-50% of the budget to text-only pretraining made images worse (gFD 48.0 to 51.6-65.1), and starting from a pretrained language model gave no gain. The same models also learn image captioning at no extra pretraining cost. Ten Halcyon checkpoints (5 sizes, 2 variants each) are on Hugging Face. Caveat: results are at 256x256 in the authors' own setup.
Source: Radical Numerics (opens in a new tab) - 09Research17 Oct 2026
GPT-6 Astra solved 49 of 50 unpublished physics research problems
TheoPhysEval, a new benchmark from 35 researchers at ICTS, TIFR, IISc, Harvard, KU Leuven, Google DeepMind and other institutions, sets 50 unpublished, research-level theoretical physics problems; a typical one is about a day's work or less for a world expert. Off-the-shelf AI agents got web search and coding tools but no human hints. GPT-6-Astra in Codex solved 49 of 50 (98%), at under $3 per problem on average (about $116 for the full run); Claude Opus 5.5 in Claude Code solved 46, Gemini 3.8 Flash 34, open-weight DeepSeek-V4.1-Flash 27 and GLM 5.3 21. Several Astra answers first graded wrong turned out to expose subtle errors in the experts' reference solutions. With web search off, Astra and Opus still solved 46 each, which the authors read as heavy memorization of the physics literature. The authors say some frontier models may already have surpassed the best human researchers at concrete calculations. Caveat: these are "turn-the-crank" problems that extend known results, not breakthroughs; each agent ran once; and agents still lack "taste" in picking problems.
Source: arXiv (TheoPhysEval) (opens in a new tab) - 10Research16 Oct 2026
Goodfire: neural networks may already be interpretable
A working hypothesis: in preliminary audits ~90% of Qwen3-32B neuron poles have one main job, so understanding AI from the inside may be far more tractable than assumed.
Source: Goodfire (opens in a new tab) - 11Research16 Oct 2026
OpenAI opens GPT-Rosalind to US university labs
Eligible U.S. academic labs get up to 30 free seats for up to 12 months and trusted access to OpenAI's life sciences model, lowering the cost of AI for basic biology research.
Source: OpenAI (opens in a new tab) - 12Research16 Oct 2026
CERN and Mistral plan a Large Physics Model
CERN and Mistral signed a statement of intent to build the first foundation model for collider physics, trained on billions of simulated LHC collisions, to learn across a reconstruction chain that today relies on specialised algorithms.
Source: CERN (opens in a new tab) - 13Research16 Oct 2026
Google Deep Review: public AI that stress-tests technical docs
Google Research launched Deep Review publicly after 16,000+ conference-submission pilots; 2 free reviews/week, PDFs/plain text, uploads not used for training; Google cites 74k+ AI-conference submissions this year.
Source: Google Research (opens in a new tab) - 14Research16 Oct 2026
Sakana open-sources The Agentic Scientist for AI-run research
Sakana AI's next AI Scientist orchestrates coding agents with rubrics, shared memory, ablations and an evidence map; a co-discovered paper was accepted as an ALIFE 2026 oral, and the code is open source.
Source: Sakana AI (opens in a new tab) - 15Research14 Oct 2026
Modal trained RL across 6 GPU types — and it stayed stable
Mixed-precision, mixed-accelerator samplers (B200/B300 NVFP4, H100/H200 FP8+BF16, A100/RTX PRO BF16) with off-policy correction kept GRPO stable; SWE-bench Pro V2 on par with homogeneous B200 despite 4.7x larger KL; ~15% lower sampler cost per update.
Source: Modal (opens in a new tab)