Blog

Beyond Safe Models: Why AI Governance Must Tackle Unsafe Ecosystems

/ ,

AI governance must shift focus from just ensuring model safety to addressing the risks of unsafe deployment ecosystems, which are influenced by institutional contexts, conflicting incentives, and inadequate oversight. While initiatives like the EU AI Act emphasize technical compliance, they often ignore the broader environment affecting AI use, leading to harmful outcomes like discrimination and misinformation. Effective governance requires assessing deployment contexts, aligning institutional incentives, ensuring accountability, and establishing adaptive oversight to manage emerging risks, ultimately recognizing that AI's dangers stem from both its operation and the settings it inhabits.

https://www.techpolicy.press/beyond-safe-models-why-ai-governance-must-tackle-unsafe-ecosystems/

A Guide on Becoming a Chief Information Security Officer

/ ,

Become a CISO to lead cybersecurity, manage risks, and protect data. Responsibilities include strategic oversight, team management, incident response, and aligning security with business goals. Key skills: cybersecurity expertise, risk management, communication, leadership, and project management. Start with a tech-related degree, gain experience, earn certifications (like CISSP, CISM), and transition to leadership roles. Continuous learning and networking are essential for success in this evolving field. Average salary: ~$309,000 annually.

https://www.techloy.com/a-guide-on-becoming-a-chief-information-security-officer/

What if the EU Was Really Serious About AI?

/ , ,

EU's AI strategy lags behind the US and China. To become competitive, it should:

  1. Infrastructure: Increase investment in cloud computing and partnering with US tech.
  2. Data: Simplify data regulations, enhance open data access, and incentivize data sharing.
  3. AI Adoption: Set ambitious AI targets and focus on outcomes in public contracts.
  4. Skills and Talent: Fund AI academic positions and pivot education programs toward AI skills.
  5. (De)Regulation: Streamline regulations for ease of use while ensuring safety.

Addressing defense AI and promoting global leadership in open-source AI is vital. Europe has the resources; bold actions are required to catch up.

https://cepa.org/article/what-if-the-eu-was-really-serious-about-ai/

R&G Tech Studio: Navigating AI Literacy—Understanding the EU AI Act

/ ,

R&G Tech Studio podcast discusses the EU AI Act, focusing on AI literacy requirements for organizations. Hosts Rohan Massey and Edward Machin explain the broad definition of AI systems under the Act and emphasize the importance of AI literacy for both providers and deployers, irrespective of risk categories. Organizations must tailor AI literacy training based on contextual needs, employee roles, and resources. They suggest that companies begin developing AI literacy strategies now, despite limited guidance, to ensure compliance and effectively manage AI-related risks.

https://www.ropesgray.com/en/insights/podcasts/2025/04/rg-tech-studio-navigating-ai-literacy-understanding-the-eu-ai-act#page=1

Managing the Many We’s of IT

/ ,

CIOs must understand the diverse identities within their IT organizations, recognizing that the “we” in IT includes various groups with differing values and engagement levels. Successful CIOs need to be transparent about their own identity and foster engagement by ensuring employees have effective tools, reducing unnecessary work, and managing vendor relationships. By broadening the concept of “we,” CIOs can enhance collaboration and collective responsibility for IT value creation and security across their organizations.

https://www.cio.com/article/3968848/managing-the-many-wes-of-it.html

EU Moves to Clarify AI Act Scope for gen-AI

/ , ,

EU proposes thresholds for computational resources to clarify compliance for general-purpose AI (GPAI) models under the AI Act effective August 2025. The guidelines, subject to industry feedback via a survey, aim to establish when AI models become subject to regulatory requirements. Key points include defining GPAI models based on compute use (>= 10^22 FLOP), obligations for record-keeping, copyright policies, and potential compliance benefits for signatories to a forthcoming code of practice. Critics argue reliance on FLOP is flawed as it may inadequately reflect model capabilities and risks. Moreover, modifications over certain compute thresholds may elevate compliance burdens.

https://www.pinsentmasons.com/out-law/news/eu-clarify-ai-act-scope-gen-ai

Scroll to Top