The AI logins went out. Why did the team stop using them?

The AI logins go out and the first week goes well. After that, use of the new tools can start to drop. Bringing the team back is a different job from choosing the tool, and it needs its own plan.

Firms are not alone in finding a gap between intention and practice. The National AI Centre’s Guidance for AI adoption reports the findings of the 2025 Responsible AI Index. It found that 78% of respondents agreed with ethical AI performance statements, while only 29% had implemented relevant responsible AI practices. Confidence in responsible AI also declined among organisations with 20 to 99 employees.

The guidance, published in October 2025, sets out six practices for responsible AI adoption. Much of what firms run into after rollout sits under the first one.

The first practice in the guidance is deciding who is accountable. Its explanation is short: “Accountability is the first step to using AI responsibly.”

The guidance also names what happens when nobody owns the tool: “AI complexity can create gaps where no one takes clear responsibility for outcomes.” In a firm, that can look like a tool everyone has a login for and nobody is responsible for.

Paragraph 1.1.1 asks organisations to name who is accountable for AI, write it down and make sure everyone knows. Paragraph 1.1.2 adds that each accountable person needs defined competencies and the authority to act.

Paragraph 1.4.3 names a gap that can open up after rollout. It asks organisations to identify and address “any gaps between leadership expectations and staff understanding of how to develop and deploy AI safely and responsibly.”

Section 1.3 of the guidance covers AI literacy and training. It links training directly to adoption, saying it builds confidence and makes sure accountable people have the capabilities their roles need.

The guidance treats training as ongoing work. Paragraph 1.3.1 says: “Regularly check skills are up-to-date as AI development and deployment evolves.” It also says anyone using AI should be trained to understand each system’s limits and when a person needs to step in.

Guardrails sit alongside the training. The guidance calls for privacy policies that cover how AI systems collect, use and disclose personal information, so teams can comply with the Australian Privacy Principles.

The guidance also expects firms to learn from what goes wrong. It says: “Reporting near-misses and corrective measures is good practice.” A rule added after a near-miss is that kind of corrective measure.

The guidance leaves the choice of owner to each organisation, as long as that person has the skills and authority to act.

For a firm, that choice shapes how adoption runs through the next year. The panel makes a 2027 call on where AI skills should sit in a firm, and closes with a 90-day action firms can take back to their own team.

The guidance sets out what responsible adoption involves. Getting a team from login to habit is the part each firm has to work out for itself.

Teach your team how to use AI productively and securely

Gus McLennan is Managing Director of AI Dojo, an AI tool built for accounting firms. James Rose is Co-founder of Content Snare, which firms use to collect documents and information from clients. With practitioners Beau Gaudron and Travis Bacon, they talk through what happens after an AI rollout. Their session covers the first 30 days, how to win the team back and the guardrails that keep client data safe. It is a chance to learn more about getting a whole team to adopt AI securely.

Wednesday 11 November 2026, 3:00pm AEDT

Register for the session

Part of The AI in Practice Summit, 11 and 12 November 2026. Free, online, every session recorded.