US intel flags AI — but skips election threat
The US intelligence community elevated artificial intelligence to a top-tier global threat in its latest assessment, yet the report notably omitted direct findings on foreign threats to the November 2026 congressional elections — DNI Tulsi Gabbard said no active foreign threat has been identified. Governments and election bodies aren’t waiting: South Africa’s IEC rolled out a ‘multifaceted shield’ against an ‘AI disinformation hurricane,’ 82% of journalists now use AI in newsrooms, and policymakers are pushing patchwork AI bills as critics warn heavy-handed mandates could backfire. (defenseone.com, nextgov.com, reuters.com, dailymaverick.co.za, manilatimes.net, rstreet.org
The unclassified 2026 Annual Threat Assessment contains a discrete “AI” section on page 12 and notes it was drafted using intelligence available as of March 14, 2026. (intelligence.senate.gov) At a March 18 Senate Intelligence Committee hearing, Vice Chair Sen. Mark Warner pointed out that this year’s assessment omitted election-influence findings for the first time since 2017 and pressed DNI Tulsi Gabbard directly on whether foreign threats to the midterms exist. (politico.com) Gabbard replied that “the intelligence community has been and continues to remain focused on any collection and intelligence that shows a potential foreign threat,” prompting Warner to push back with “so far, there has been none,” while lawmakers also faulted the agencies for declining to provide briefings on election threats. (nextgov.com, reuters.com) South Africa’s IEC outlined a “multifaceted shield” strategy that includes “radical transparency,” pre-bunking campaigns, strengthened media partnerships, and considering the EU’s AI Act as a blueprint for categorizing high‑risk systems ahead of its 2026 local elections. (dailymaverick.co.za) Muck Rack’s State of Journalism 2026 surveyed 1,044 journalists between Jan. 30 and March 2 (897 valid responses) and found 82% now use at least one AI tool, with 47% reporting ChatGPT use and 22% using Gemini, while 26% listed “unchecked AI” as a top concern. (markets.financialcontent.com) Policy groups warn that state-level AI bills are proliferating—R Street says more than 1,500 AI-related bills are circulating and highlights a list of especially problematic proposals dubbed the “AI Terrible Ten,” even as cybersecurity firms report operational impacts: CrowdStrike found AI-enabled attacks surged 89% and average breakout time fell to 29 minutes in its 2026 threat report. (rstreet.org; crowdstrike.com)
