AI This Week: State AI Laws Kick In and Agents Enter Webex
Two state AI laws moved from the statute books into active compliance territory this past week. On October 1, Connecticut’s comprehensive AI regulation took effect. A day earlier, California’s governor signed the No Robo Bosses Act — giving California employers until July 2027 to adjust how they use AI for employment decisions. While regulators raised the floor on governance, Cisco’s WebexOne conference in Austin signaled what AI adoption now looks like at the workplace level: AI agents that join your meetings and take action from them.
Connecticut’s AI Regulation Is Now Live — and It May Apply to Your Business
Connecticut’s Artificial Intelligence Responsibility and Transparency Act (Public Act 26-15, also called the CART Act) took effect October 1, 2026. Signed into law this past June, it imposes risk management and transparency obligations on businesses that deploy “high-risk AI systems” — defined broadly enough to cover AI used in hiring, lending, insurance underwriting, housing access, and more.
What the law requires right now, per Ropes Gray’s client alert on the CART Act: deployers must implement written risk management programs, conduct impact assessments before deployment and annually afterward, and give consumers notice when high-risk AI is used in decisions affecting them. The employment-specific provisions — covering hiring and performance management AI — don’t kick in until October 1, 2027, but Faegre Drinker’s analysis advises mapping your AI use cases now rather than scrambling next fall.
The obligations don’t require you to be incorporated in Connecticut. They apply based on where your affected customers or workers are located — which means any business serving Connecticut residents could be in scope.
What this means for your business: If you use AI to screen applicants, score leads, set credit terms, or make any decision that materially affects a customer in Connecticut, you have an active compliance obligation today. A short audit of your AI systems and their use cases is the right first step — and far cheaper than a retroactive one. Commercial real estate and multifamily operators using AI for tenant screening are high on the list of businesses that should treat this as immediate.
California Just Banned AI-Only Firing Decisions
On September 30, California Governor Gavin Newsom signed SB 947, known as the No Robo Bosses Act. The Next Web reported on the signing as part of a batch of 13 AI bills Newsom approved the same day. Employment law firm Fisher Phillips published a five-step employer compliance guide within 24 hours of the signing.
The law takes effect July 1, 2027, giving California employers roughly eight months to prepare. Its core rule: no employer may rely solely on an automated decision system to discipline or terminate a worker. A human reviewer must check the system’s output against independent evidence — manager evaluations, peer reviews, documented performance data — and simply clicking “approve” on an AI recommendation does not count. Newsom vetoed a nearly identical bill in 2025. This version removed gig-worker protections, but the core prohibition on fully automated adverse employment decisions survived.
What this means for your business: If you use AI tools in your HR, scheduling, or workforce management stack that can generate performance ratings, flag attendance patterns, or recommend disciplinary action, California’s law requires a human to be genuinely in the loop — not rubber-stamping the output. Freight logistics and specialty contractors with hourly workers managed partly through AI systems should review those workflows now, before the July 2027 enforcement date arrives.
Cisco Brought AI Agents Into Webex — and They Can Attend Your Meetings
At its WebexOne 2026 conference in Austin on October 7, Cisco announced a significant expansion of AI agent capabilities for the Webex collaboration platform. RTTNews covered the event, and UCToday reported on the Claude integration specifically: Cisco is building Claude Managed Agents into Webex, using Anthropic’s managed agent infrastructure rather than a simple chatbot plugin.
The announced capability allows AI agents to be mentioned directly into a Webex space, meeting, or call — the way you would @mention a colleague — then execute multi-step work across Webex and connected third-party applications. The agent participates in context, not just passively transcribing. Most features are slated for early 2027, so this is an announced roadmap, not a shipping product today. But the architecture matters: it signals where workplace AI is heading — away from sidebar tools you context-switch to and toward agents that are already present in the meeting room.
What this means for your business: If your team runs on Webex and you’ve been watching AI demos without seeing something that fits your actual workflow, the WebexOne model is worth tracking. The “AI agent in your meeting” approach removes the manual step of routing outputs between tools — action items can trigger directly from the conversation. Construction and energy services firms that coordinate large teams across multi-party Webex calls stand to benefit most from agents that can capture decisions and initiate follow-up work without a human routing step.
The takeaway
State-level AI regulation isn’t coming — it’s here. Connecticut’s CART Act is live, California’s employer rules are eight months from enforcement, and the pattern is clear: how you govern your AI use will be a legal requirement, not just a best practice. At the same time, tools like Cisco’s Webex agent integration show that the practical case for AI in the workplace keeps advancing fast. The businesses that will navigate this well are the ones building governance habits now, while the tools are still being introduced. If you want help understanding what any of this means for your specific operation, reach out.