Organizations are deploying AI at breakneck speed—so rapidly, in fact, that foundational safeguards like governance and access controls are being sidelined. The 2025 IBM Cost of a Data Breach Report, based on data from 600 breached companies, finds that 13% of organizations have suffered breaches involving AI systems, with 97% of those lacking basic AI access controls. IBM refers to this trend as “do‑it‑now AI adoption,” where businesses prioritize quick implementation over security.  The consequences are stark: systems deployed without oversight are more likely to be breached—and when breaches occur, they’re more costly. One emerging danger is “shadow AI”—the widespread use of AI tools by staff without IT approval. The report reveals that organizations facing breaches linked to shadow AI incurred about $670,000 more in costs than those without such unauthorized use.  Furthermore, 20% of surveyed organizations reported such breaches, yet only 37% had policies to manage or detect shadow AI. Despite these risks, companies that integrate AI and automation into their security operations are finding significant benefits. On average, such firms reduced breach costs by around $1.9 million and shortened incident response timelines by 80 days.  IBM’s Vice President of Data Security, Suja Viswesan, emphasized that this mismatch between rapid AI deployment and weak security infrastructure is creating critical vulnerabilities—essentially turning AI into a high-value target for attackers. Cybercriminals are increasingly weaponizing AI as well. A notable 16% of breaches now involve attackers using AI—frequently in phishing or deepfake impersonation campaigns—illustrating that AI is both a risk and a defensive asset.  On the cost front, global average data breach expenses have decreased slightly, falling to $4.44 million, partly due to faster containment via AI-enhanced response tools. However, U.S. breach costs soared to a record $10.22 million—underscoring how inconsistent security practices can dramatically affect financial outcomes.  IBM calls for organizations to build governance, compliance, and security into every step of AI adoption—not after deployment. Without policies, oversight, and access controls embedded from the start, the rapid embrace of AI could compromise trust, safety, and financial stability in the long run.

TikTok and EU regulation: Legal challenges and cross-jurisdictional insights

New study argues expanding copyright control over AI training data could harm innovation and public interest.

The grave risk that AI-based decision support systems (AI DSS) for target selection will actually exacerbate civilian suffering.

This milestone marks a significant advancement in international AI governance and the safety of advanced AI systems.

Building Trust in #AISystems: Key Principles for Ethical and Reliable AI Which of the five ethical AI principles involves making AI systems understandable and transparent to users and stakeholders? 1....