From Safety to Security: Governing Adversarial Use of AI скачать в хорошем качестве

From Safety to Security: Governing Adversarial Use of AI

The Perilous Race: Navigating AI's Adversarial Frontiers and the Crisis of Collective Governance The panel, moderated by Océane Herrero (Tech Reporter, POLITICO), explored the shift from "AI safety" (preventing unintended systemic consequences) to "AI security," highlighting risks from malicious actors using AI for cyberattacks, disinformation, and potential weaponization. The discussion emphasized that technological acceleration may outpace ethical governance and global oversight. Accelerating Threats Yoshua Bengio (Co-President, LawZero), lead author of the first international AI risk report following the Bletchley Park summit, detailed how AI amplifies threats—from deepfakes and fraud to targeted influence campaigns and cybersecurity breaches. Bengio warned of "excessive concentration of power" in a few corporations and states, destabilizing governance and economic balance. Nicolas Butts (Director, Global Cybersecurity and AI Policy, Microsoft) noted a five-fold increase in AI-driven operations by nation-states, including digital cloning, targeted propaganda, and a 4.5x surge in ransomware effectiveness. He stressed that vulnerabilities persist despite basic cybercriminal tactics being effective, signaling structural gaps in cybersecurity defenses. Governance Gaps and Proactive Measures Jacinda Ardern (Patron, Christchurch Call) recounted the Christchurch attack, where AI-enabled live-stream amplification revealed inadequate safeguards. The Christchurch Call—a multi-stakeholder initiative—developed crisis response protocols and algorithmic transparency tools, showing collective action’s potential. Ardern emphasized the challenge of establishing AI guardrails proactively, warning that commercial pressures often deter regulation. Robin Geiss (Director, UNIDIR) highlighted "machine warfare," where autonomous AI systems accelerate conflict, compromise ethical oversight, and creep into nuclear and chemical weapon domains. He stressed the "speed bump" dilemma: nations hesitate to self-regulate for fear of losing strategic advantage. Socio-Technical Imperative Vilas Dhar (President, Patrick J. McGovern Foundation) argued AI risks are as much social as technical. Beyond data bias and algorithmic errors, building resilient institutions, accountability, and multilateral governance is essential. Dhar challenged the "AI arms race" narrative as politically convenient but dangerous, advocating investment in human-aligned AI. Initiatives like the OECD AI Incident Monitor and the nascent UN International Scientific Panel are steps toward democratized decision-making. He stressed empowering the Global South as an equal partner, not merely a consumer or testing ground for AI technologies. Strategic Takeaway AI governance requires more than technical solutions—it demands rethinking global power structures and committing to human-centric principles. The competitive race for technological dominance incentivizes recklessness, creating governance gaps and escalating adversarial AI uses. The panel called for a "coalition for humanity," engaging governments, industry, civil society, and the Global South to proactively develop ethical guardrails. By redefining AI as a shared, human-aligned endeavor, the international community can move from reactive crisis management to shaping a future where AI enhances security and human dignity rather than threatening it. Follow us on social media • 𝗧𝘄𝗶𝘁𝘁𝗲𝗿 -   / parispeaceforum   • 𝗟𝗶𝗻𝗸𝗲𝗱𝗜𝗻 -   / parispeaceforum   • 𝗙𝗮𝗰𝗲𝗯𝗼𝗼𝗸 -   / parispeaceforum   • 𝗜𝗻𝘀𝘁𝗮𝗴𝗿𝗮𝗺 -   / parispeaceforum   • 𝗧𝗶𝗸𝗧𝗼𝗸 -   / parispeaceforum_