Heidelberg AI News

All stories

119 stories

Clear

Wednesday 21 October

Tuesday 20 October

  • OpenAI

    OfficialOpenAI

    Terminal X matches senior investor judgment 88.6% with GPT-6.1

    OpenAI's Terminal X includes an Eval Agent that grades AI outputs against a 54-criterion rubric. Using GPT-6.1 Sol, it matches senior investor judgment 88.6% of the time and catches 25% more deviances from a fund's mandate.

  • OpenAI

    OfficialOpenAI

    OpenAI acquires Orchestra Bio

    OpenAI is acquiring Orchestra Bio to bring AI-powered operations to life sciences R&D, helping biopharma teams accelerate drug development and advance new treatments.

Monday 19 October

Friday 16 October

Thursday 15 October

Wednesday 14 October

  • OfficialDocker

    Harbor + Docker Sandboxes: Run Safe, Reproducible Agent Evals

    Docker's Harbor can now run every eval task in Docker Sandboxes, which are MicroVM environments for running agents like Claude Code, Codex, and Cursor safely. With Sandboxes you can define custom network policies for fair, reproducible evals, and you can run evals locally or in the cloud, starting free on Docker Sandboxes on local surfaces and scaling up.

  • OfficialGitHub

    HydraFusion updates and new Max option

    GitHub announced updates to HydraFusion, including an expanded model pool and improved orchestration workflows, and introduced a new option called HydraFusion Max. Both HydraFusion and HydraFusion Max are available across VS Code, the GitHub Copilot app, and GitHub Copilot.

  • OfficialGitHub

    GitHub Actions workflow lockfiles in public preview

    GitHub Actions workflow dependency lockfiles are now in public preview for GitHub Enterprise Cloud and GitHub Enterprise Cloud with data residency. The announcement notes that tags and branches can move after a review.

  • OpenAI

    OfficialOpenAI

    The EU's next chapter in AI is agentic

    OpenAI analysis shows agentic AI adoption growing across EU enterprises and small businesses, identifying where deeper use could unlock more value.

Tuesday 13 October

  • OfficialGitHub

    Gemini 4 Argon in GitHub Copilot

    Gemini 4 Argon, Google's newest frontier model, is now generally available in GitHub Copilot. It is designed for complex coding, agentic, and reasoning tasks.

  • Google DeepMind/Google Research

    OfficialGoogle DeepMind/Google Research

    Gemini 4 Argon: Built to solve your hardest problems

    Google DeepMind/Google Research announced Gemini 4 Argon, its latest frontier model, giving developers, enterprises, scientists, and consumers access to it.

  • OfficialGitHub

    Search and filter for budgets now available

    GitHub's 'Budgets and alerts' page in billing settings now displays all of a user's budgets in a single list that can be searched and filtered, making it easier to manage budgets.

  • OpenAI

    OfficialOpenAI

    Higgsfield takes AI video from 1 minute to over 10

    Higgsfield, an AI video tool, now lets creators make videos over 10 minutes long, up from about 1 minute. Per OpenAI, the update adds longer videos via GPT-6 Astra, more precise visuals with GPT-Image-2.5, and a path toward voice filmmaking with GPT-Live.

  • OfficialGitHub

    Client telemetry user attribution and configurable capture

    GitHub announced that enterprise administrators can now use telemetry in managed settings to attribute GitHub Copilot activity to a user without collecting their prompts, responses, or tool content. The new telemetry.capture.identity setting adds three attributes to enable this attribution.

Saturday 10 October

Friday 9 October

Thursday 8 October

  • OfficialGitHub

    Copilot code review: New organization billing options and controls

    GitHub added new billing and license controls for Copilot code review admins. Organization owners can now bill Copilot code reviews from members with a Copilot license to the organization's billing, along with new license controls.

  • OpenAI

    OfficialOpenAI

    Disrupting AI-enabled "false front" operations

    OpenAI announced that it disrupted two AI-enabled influence operations. These operations used false-front journalists and a think tank to spread geopolitical messaging.

Wednesday 7 October

  • OfficialGitHub

    Claude Haiku 5.5 in GitHub Copilot

    Claude Haiku 5.5, Anthropic's lightweight model, is now generally available in GitHub Copilot. It is designed for fast, high-volume work such as subagents, quick edits, and terminal tasks, according to The GitHub Blog.

  • OfficialGitHub

    Local sandboxing for GitHub Copilot now generally available

    GitHub Copilot's local sandboxing feature is now generally available across GitHub Copilot CLI, the GitHub Copilot app, and VS Code sessions using Agent Host. Local sandboxes give developers a secure environment.