82搜索为您找到约 3,493 条结果
www.anthropic.com
网页
综合
Mapping AI-enabled cyber threats \ Anthropic
… a hypothetical attack is likely to succeed. Under this model, if any one factor is zero, the overall risk collapses to zero, because a missing ingredient means the attack will not succeed. …
www.anthropic.com
网页
综合
Testing and mitigating elections-related risks \ Anthropic
…n the case of subjective questions) answers, note the presence of any safety interventions (e.g., refusing to answer a harmful question), and detail their qualitative findings. For example, …
www.anthropic.com
网页
综合
Disrupting an AI-orchestrated cyber espionage campaign \ Anthropic
… human operators chose the relevant targets (for example, the company or government agency to be infiltrated). They then developed an attack framework—a system built to autonomously compromi…
www.anthropic.com
网页
综合
Claude SWE-Bench performance \ Anthropic
…folding, even when using the same underlying AI model. There are many other benchmarks for the coding abilities of Large Language Models, but SWE-bench has gained in popularity for several r…
www.anthropic.com
网页
综合
Our updated Responsible Scaling Policy \ Anthropic
…an scheduled or a lack of clarity on how and where we should note any changes to our placeholder evaluations. We also flagged some evaluations where we may have been able to elicit slightly …
www.anthropic.com
网页
综合
Measuring LLMs’ ability to develop exploits \ Anthropic
…given vulnerabilities, but only models since Claude Opus 4.6 make any progress in developing primitives inside the V8 sandbox. Escaping the V8 sandbox, going from T3 to T2, is the next capab…
www.anthropic.com
网页
综合
Claude Mythos Preview's cybersecurity capabilities \ Anthropic
…ulnerabilities it finds are often subtle or difficult to detect. Many of them are ten or twenty years old, with the oldest we have found so far being a now-patched 27-year-old bug in OpenBSD…
www.anthropic.com
网页
综合
AI agents find smart contract exploits \ Anthropic
…5 in simulation against 2,849 recently deployed contracts without any known vulnerabilities. Both agents uncovered two novel zero-day vulnerabilities and produced exploits worth $3,694, with…
deepmind.google
网页
综合
AlphaEarth Foundations helps map our planet in unprecedented detail — Google DeepMind
…aking non-uniformly sampled frames from a video sequence to index any position in time. This helps the model create a continuous view of the location, while explaining numerous measurements.…
deepmind.google
网页
综合
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms — Google DeepMind
…ical problems, showing incredible promise for application across many areas. In 2023, we showed for the first time that large language models can generate functions written in computer code …
第 339 / 350 页