Anthropic, AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks join forces to use Claude Mythos Preview for defensive cybersecurity at scale.
OpenAI adds WebSocket support to the Responses API, cutting end-to-end latency for tool-heavy agentic workflows and simplifying long-running session handling.
OpenAI launches workspace agents in ChatGPT Business and Enterprise, letting teams deploy persistent agents that operate across connected apps, files, and tools with admin-managed permissions.
OpenAI introduces Codex Transformation Partners, a global program to help large enterprises adopt Codex at scale alongside hands-on rollout support and integration tooling.
GitHub announces structural changes to the Copilot Free, Pro, and Pro+ tiers for individual users, including updates to premium-request quotas and model availability.
Claude Design is a new Anthropic Labs product that lets you collaborate with Claude to create polished visual work — designs, prototypes, slides, one-pagers, and more.
OpenAI expands Codex's reach across new surfaces and workflows, positioning it as the default agentic coding interface for ChatGPT Business, Enterprise, and API users.
Anthropic's latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most.
OpenAI outlines its strategy and partnerships for AI-powered cybersecurity defense — a counterpart announcement to Anthropic's Project Glasswing.
Public preview lets you remotely view, steer, and respond to active Copilot CLI sessions from github.com or GitHub Mobile, bringing parity with the cloud-agent experience.
Copilot cloud agent can now detect, analyze, and resolve merge conflicts in pull requests with a three-click flow, drafting a merge commit you can review before accepting.
GitHub is pausing new Copilot Pro free trials while the team reworks the individual plan structure, ahead of broader changes to Copilot for individuals.
Organization admins and security managers can now jump into a Copilot experience directly from secret risk assessment or Code Security risk assessment results to get contextual explanations and guided next steps.
VS Code's April 2026 release introduces the VS Code Agents companion app preview, enhanced integrated browser capabilities for agents, and new terminal tools for background command interaction.
The combined March/April Copilot changelog covers releases v1.111 through v1.115, including Autopilot for fully autonomous agent sessions, integrated browser debugging, image and video support in chat, and a new unified customizations editor.
Copilot cloud agent now works beyond pull-request workflows on GitHub Mobile, letting you research codebases, generate implementation plans, and iterate on branches from your phone.
The Copilot usage metrics API now includes two new metrics tracking Copilot code review activity: total merged PRs reviewed by Copilot and their median time to merge.
OpenAI CRO Denise Dresser outlines the company's enterprise strategy: enterprise now exceeds 40% of revenue, Codex has 3M weekly users, and APIs process 15B tokens per minute.
You can now assign Dependabot security alerts directly to Copilot, Claude, or Codex. Each agent analyzes the vulnerability and opens a draft PR with a proposed fix.
GitHub Copilot CLI can now connect to any OpenAI-compatible provider or run fully local models like Ollama and vLLM, enabling air-gapped development workflows.
Anthropic signs a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity coming online in 2027, as run-rate revenue surpasses $30 billion.
Organization admins can now manage the Copilot cloud agent firewall across all repositories, controlling internet access and protecting against prompt injection and data exfiltration at scale.
GitHub has officially deprecated the GPT-5.1 Codex family of models from Copilot, completing the transition to GPT-5.4 across all coding agent surfaces.
Organization admins can now set default runners and lock runner settings across all repositories for Copilot cloud agent, removing the need for per-repo copilot-setup-steps.yml configuration.
Copilot cloud agent now signs every commit it makes, appearing as Verified on GitHub. This also enables the agent to work in repositories with "Require signed commits" branch protection enabled.
Copilot Business and Enterprise admins can now set default custom instructions that guide Copilot's behavior across all repositories in their organization, including Chat, code review, and cloud agent.
The GitHub Copilot SDK ships in five languages, exposing the same agentic runtime that powers Copilot cloud agent and Copilot CLI so teams can embed Copilot's capabilities directly into their own applications.
OpenAI introduces Codex-only seats with pay-as-you-go pricing for ChatGPT Business and Enterprise workspaces, plus lowers the annual Business seat price from $25 to $20.
Copilot cloud agent is no longer limited to pull-request workflows. It can now work on branches without creating PRs, generate implementation plans for review, and conduct deep research sessions grounded in your repository context.
OpenAI details the shell tool, hosted containers, skills, compaction, and network controls it added to the Responses API so agents can run longer, safer workflows.
GitHub added a file explorer directly to Copilot on the web, letting you inspect repository structure, preview files, and attach them as chat references in context.
OpenAI argues that prompt injection increasingly behaves like social engineering and says secure agents need constrained actions, confirmation gates, and safer outbound behavior.
VS Code's March 2026 release (1.111) added session-level permissions for agent autonomy, an Autopilot preview, agent-scoped hooks, and debug event snapshots for troubleshooting.
OpenAI says GPT-5.4 combines GPT-5.3-Codex coding ability with stronger tool use, computer use, and professional-task performance across ChatGPT, the API, and Codex.
GitHub moved Copilot code review onto an agentic tool-calling architecture so reviews can gather repository context on demand and produce fewer, higher-signal comments.
GitHub rolled GPT-5.4 into Copilot across IDEs, github.com, mobile, CLI, and the coding agent, with plan-level controls for business and enterprise admins.
Anthropic's March 2026 recorded session is a practical onboarding walkthrough for Claude Code, covering setup, persistent project context, prompting structure, and an end-to-end live task demo.
Anthropic's recorded February 2026 event focuses on how Shortcut and Hex are using Claude Opus 4.6 in production, including live demos, prompting strategy, and lessons learned.
Anthropic positions Sonnet 4.6 as a full-stack upgrade for coding, computer use, long-context work, and agent planning, while keeping Sonnet pricing and making it the new default across Claude surfaces.
Anthropic researcher Nicholas Carlini documents what 16 parallel Claude agents could achieve in a shared codebase, and where the approach still breaks down.
OpenAI's DevDay session focuses on how its own engineers are using Codex across local and cloud workflows to write, refactor, and merge code faster.
GitHub demonstrates a complete end-to-end development workflow using Copilot's agentic capabilities, from planning through implementation and testing.
Anthropic demonstrates best practices for prompting AI coding agents, including how to structure instructions and context for Claude to handle complex multi-step development tasks.
Anthropic showcases practical production coding workflows with Claude, covering real-world scenarios from feature development to debugging in production codebases.
Learn the most effective prompting techniques for AI coding assistants. Covers prompt patterns, context management, and strategies for getting better code suggestions.
An in-depth walkthrough of VS Code's agent mode capabilities, demonstrating how AI-powered agents can handle complex development tasks directly within the editor.