OpenAI expands Daybreak Cyber Partner Program with Accenture, IBM, Palo Alto Networks, and CrowdStrike

Key Takeaways

OpenAI shares cyber models with trusted partners

OpenAI is expanding its Daybreak Cyber Partner Program to bring its advanced cyber models to security partners around the world. The program includes leading security and services partners such as Accenture, IBM, Palo Alto Networks, and CrowdStrike. These partners will help more defenders find serious vulnerabilities, validate which ones matter, and fix them faster. The program aims to close the growing defense gap by providing a practical way for organizations to benefit from frontier cyber models.

OpenAI advances cyber defense with partner program

OpenAI's Daybreak Cyber Partner Program makes frontier cyber models available to security partners, allowing organizations to access advanced defensive capabilities without having to build and run their own specialized cyber AI programs. The program includes partners such as Accenture, IBM, Palo Alto Networks, and CrowdStrike, bringing deep security expertise and established relationships with organizations worldwide.

OpenAI expands Daybreak for cyber defense

OpenAI is expanding its Daybreak program with two access tiers designed to give approved defenders the right capabilities for their work. Daybreak Blue provides access to frontier general-purpose models, while Daybreak Red is for more specialized, closely governed work. OpenAI also introduces GPT-5.6-Cyber, a cybersecurity-specific model available through Daybreak Red.

AI hijacking attack uses poisoned logs

Researchers demonstrate a novel AI hijacking attack that relies on tools trusted by AI agents to deliver malicious instructions. The attack targets highly trusted platforms such as Cloudflare, Datadog, and Sentry, which are used by nearly half of Fortune 500 companies. The compromised AI agent can be tricked into executing code, stealing credentials, and turning into an insider that vouches for the attacker.

GhostJacking exposes AI agent vulnerabilities

Researchers show how attackers can use security alerts and blocked events to manipulate and hijack AI agents. The attacks exploit the fact that AI agents consume data from trusted systems and can take actions based on that information. The technique was demonstrated on Cloudflare, Datadog, and Sentry, but the underlying attack pattern is not limited to any single platform.

Google DeepMind faces challenges in AI race

Google DeepMind is struggling to retain its technical edge, keep hold of top researchers, and contain an employee revolt over its defense work. The company's CEO, Demis Hassabis, is stepping down, and a new leadership structure is being put in place. Google DeepMind is facing delays in model releases and is trying to catch up with rivals such as OpenAI and Anthropic.

Tenable One launches AI security capabilities

Tenable One introduces new AI security capabilities, including visibility and governance for Google Gemini, policy enforcement, and detection of malicious activity. The platform also offers operationalized remediation, enabling organizations to respond quickly to AI policy violations.

New York AI startup expands to Connecticut

Queen One, a Brooklyn-based e-commerce technology startup, plans to expand its Connecticut workforce after securing an investment of up to $2 million from Connecticut Innovations. The company develops AI software for online retailers and aims to establish recruiting relationships with Yale University and UConn.

AWS integrates with OpenAI Codex and Anthropic Claude Code

AWS announces that it is embedding its Continuum security tool into OpenAI Codex, Anthropic Claude Code, and Kiro as it expands Security Hub Extended to secure AI-generated code and software supply chains. The move aims to position AWS as the default security control plane for enterprise software development in the AI era.

AI resilience strategy drives Veeam's recovery focus

Veeam emphasizes the importance of AI resilience in its data recovery and management strategies. The company highlights the need for organizations to prepare for AI-related disruptions and ensure data resilience in the face of increasing cyber threats.

AI investments lead to decreased training budgets

Research finds that companies are allocating more funds to AI investments, resulting in decreased training budgets. Many HR executives believe that AI can replace some learning and development activities, leading to a reduction in training expenditures.

Howard University expands AI education

Howard University is launching a new certificate program in AI fundamentals, as well as expanding its curriculum at the graduate level. The courses will cover topics such as AI tools and technologies, ethics, and bias.

Sources

• Putting frontier cyber models in more trusted hands
• Putting frontier cyber models in more trusted hands
• Expanding Daybreak as the Cyber Defense Window Narrows
• ‘Ghostjacking’ Attack Uses Poisoned Logs to Turn AI Agents Bad
• "GhostJacking" Exposes Identity Governance Gaps in AI Agents
• How stalled models, missed deadlines and staff burnout led to the unraveling of Google’s DeepMind
• New AI security capabilities launched within Tenable One
• New York AI startup plans Connecticut expansion after CI investment
• AWS Continuum integrates with OpenAI Codex and Anthropic Claude Code in major AI security push
• AI resilience strategy drives Veeam’s recovery focus
• AI investments trigger decreased training budgets
• Howard University expands AI education opportunities for students