Mats
PulseAugur coverage of Mats — every cluster mentioning Mats across labs, papers, and developer communities, ranked by signal.
- 2026-08-19 hiring MATS has opened applications for its Winter 2027 research fellowship focused on AI alignment, security, and governance. source
4 day(s) with sentiment data
-
New AsyncCouple-Flow method improves spatio-temporal forecasting with missing data
Researchers have developed a new method called AsyncCouple-Flow to improve multi-modal spatio-temporal forecasting. This approach addresses challenges such as different data sampling rates, missing modalities, and error…
-
AI reasoning traces found to be easily decryptable, exposing data
Researchers have discovered that the "encrypted" reasoning traces used by major AI providers like OpenAI, Anthropic, and Google are not truly secure. These traces, intended to protect proprietary reasoning and sensitive…
-
Modder creates unreleased 16GB Nvidia RTX 3070 Ti
A modder has successfully created a functional Nvidia RTX 3070 Ti with 16GB of VRAM, a configuration that Nvidia never officially released. The modder, known as fmklab, combined an RTX 3070 PCB with an RTX 3070 Ti GPU a…
-
AI Safety Researcher Details Career Choices and Motivations
The author reflects on their career path, emphasizing a strong inclination towards technical AI safety research. They detail a progression through various roles, including PhD candidate, fellow at MATS, independent rese…
-
Scott Alexander's ACX Open Thread 448 features OpenAI Foundation jobs and AI fellowship
Scott Alexander's Astral Codex Ten is hosting an open thread for community discussion and announcements. Key updates include a call for organizers for Effective Altruism college meetups, the release of a voting guide fo…
-
New Enrich-Retrieve-Rank pipeline scales agent capability discovery
Researchers have introduced the Enrich-Retrieve-Rank (ERR) pipeline to improve capability discovery in large agent ecosystems. This method addresses the limitations of in-context routing, which struggles as the number o…
-
MATS opens Winter 2027 AI alignment fellowship applications
MATS is now accepting applications for its Winter 2027 research fellowship, focusing on AI alignment, security, and governance. The 12-week program, running from January to April 2027, offers a $6,400 monthly stipend, s…
-
Manifund seeks regrantors for 2026 funding initiatives
Manifund is seeking individuals to serve as regrantors for its 2026 funding initiatives, aiming to distribute $50,000 pots to each selected regrantor. The organization is looking for people with strong taste, dealflow, …
-
Less Wrong proposes AI safety funding modeled on startup equity markets
Less Wrong has proposed a new system for funding AI safety organizations, modeled after startup venture capital. This "impact market" would treat donations as investments in "impact equity," allowing early donors to ben…
-
New VLA frameworks advance autonomous driving perception and action planning · 9 sources tracked
Multiple research papers introduce novel frameworks for autonomous driving that integrate vision, language, and action (VLA) capabilities. MATS proposes a multi-modality, multi-task learning approach with adaptive fusio…
-
Study questions NLA usefulness due to initialization robustness
A new study has revealed that natural language autoencoders (NLAs), designed to explain LLM thought processes, are surprisingly robust to initialization errors. Researchers found that even when initialized with entirely…
-
Google DeepMind Explores Why SFT Filters Fail for LLM Safety
Google DeepMind researchers are investigating why supervised fine-tuning (SFT) filters for safety properties in language models often fail. Their analysis, focusing on Gemini and Olmo, reveals that undesirable traits li…
-
AI safety researchers urged to join fellowships over solo papers
Aspiring AI safety researchers are advised against attempting to publish their first paper as a solo author. The author suggests that collaborative environments, such as part-time research fellowships, offer a much high…
-
Scott Alexander's ACX Open Thread Covers AI Safety, Politics, and Fellowships
Scott Alexander's Astral Codex Ten open thread highlights several key topics. One discussion point revolves around surprising alumni of the Frankfurt School, including Palantir founder Alex Karp. The thread also feature…
-
Advice for Aspiring Research Managers in AI Safety
This article offers advice for individuals interested in research management (RM), particularly within the context of the Machine Assistance & Training Services (MATS) program. The author emphasizes that RM is primarily…
-
AI safety fellowships urged to offer better feedback for applicants
An AI safety researcher suggests improvements for fellowship application processes, advocating for more constructive feedback to rejected candidates. The author proposes that fellowships provide detailed recommendation …
-
MATS AI safety program demands intense work, high compute, and strong references
A retrospective on the Machine Alignment Training (MATS) program offers insights into the demanding work schedule and significant compute resources required for AI safety research. The author, a former MATS participant,…
-
AI Safety Bootcamp Oxford offers technical and generalist tracks
OAISI is organizing its fourth AI Safety Research Bootcamp (ARBOx4) in Oxford from June 28 to July 10, 2026. The program offers two tracks: a Technical Research Stream focusing on ML safety techniques and a new Generali…