AI digest: synthetic influencers, agentic banking hires, and AI policy moves

Window checked

The five things

1. Brands are using AI-generated influencers in ads without obvious labels

• A Guardian investigation found brands quietly using AI-generated influencers in social posts that can look like real customer experiences. 1
• The article says current UK advertising rules do not specifically require brands to label AI-created promotional content, while EU AI Act labeling duties for AI-generated or manipulated content start applying in August. 1
• This matters for AI companies because synthetic content is moving from novelty demos into performance marketing, where disclosure, consumer trust and ad enforcement will decide how far brands can push it. 1

2. Lloyds is hiring 300 AI specialists for agentic banking work

• Lloyds Banking Group said it plans to hire 300 tech experts by September to work on agentic AI, adding to a 1,000-person AI team that also includes retrained staff. 2
• The projects include fraud detection, internal document search, online banking that answers plain-language finance questions, and personalization for savings and investment products. 2
• The bank says generative AI added £50 million to its balance sheet last year and expects a £100 million benefit this year, even as it leaves open the possibility of future job cuts tied to AI adoption. 2

3. Google DeepMind is framing advanced agents as insider-threat risks

• Security Boulevard reported that Google DeepMind published an "AI Control Roadmap" for containing and monitoring AI agents that might bypass oversight, leak data or sabotage assigned tasks. 3
• The roadmap introduces TRAIT&R, a taxonomy modeled on MITRE ATT&CK, with threat categories that include hidden deployments, subtle manipulation of research results, model-weight exfiltration and infrastructure damage. 3
• DeepMind says it has audited roughly one million coding-agent tasks to build live monitoring for Gemini Spark, which puts agent governance closer to enterprise security operations than abstract alignment debate. 3

4. John Jumper is leaving Google DeepMind for Anthropic

• TechCrunch reported that John Jumper, the AlphaFold scientist who shared the 2024 Nobel Prize in chemistry, is leaving Google DeepMind for Anthropic after nearly nine years. 4
• Jumper said DeepMind CEO Demis Hassabis had taken a chance on him by letting him lead the AlphaFold team six months after he finished his PhD. 4
• The move gives Anthropic a high-profile AI-for-science researcher, and it follows another DeepMind-to-frontier-lab departure this week: Noam Shazeer leaving for OpenAI. 4

5. Canada’s AI and digital-policy package moves into the fall queue

• The Canadian Press reported that Ottawa introduced a national AI strategy, a children-under-16 social-media bill, and a privacy bill during the final two weeks before Parliament’s summer break. 5
• The digital safety bill would impose responsibility duties on companies behind AI chatbots, including steps to lower harmful-content risk and crisis protocols for self-harm, suicide or violence. 5
• The privacy bill would treat privacy as a fundamental right, require transparency for automated decision-making in significant decisions, and create fines that can reach $25 million or 5% of global revenue for the most serious offenses. 5