> ## Content Index
> Fetch the complete content index at: https://www.50footmedia.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The Committee Still Decides
- URL: https://www.50footmedia.com/the-committee-still-decides/
- Published: 2026-11-21T21:46:16.000Z
- Updated: 2026-11-21T21:46:16.000Z
- Description: Standard Bank Group, a South African bank, reports that an automated AI risk assessment, built with AWS, cut a manual two-week review to a report in about two hours across six assessments. The risk committee still decides. The figures are company-reported in a vendor co-written AWS post.
- Author: Tom Bedwell
- Tags: Signal, financial-services

If you sit on a bank's AI risk committee, your job is to say yes or no before a new tool reaches customers or employees. AI's first real payoff in regulated work may be getting that decision made faster, not replacing the people who make it. The AI Value Index finds a signal of that kind from Standard Bank Group, a South African bank. In an Amazon Web Services blog post dated November 20, 2026, co-written with the bank, Standard Bank and AWS describe an automated assessment that scores each generative AI use case against international AI standards and the bank's own policies.

From the committee's seat, the rules were not the problem. The bank did not need a new AI policy. It already had a model risk practice and a change risk process. The problem was the clock. Each new generative AI application still needed a risk committee's approval before it reached customers or employees, and the post says teams can now build one of those applications "in days." When that approval "depends on reviewers reading documents and testing the system by hand, governance becomes the slowest step in the pipeline."

[The AWS post](https://aws.amazon.com/blogs/machine-learning/how-standard-bank-group-automates-ai-risk-assessment-for-generative-ai/?ref=50footmedia.com) describes what the bank, working with AWS Professional Services, did next. Evidence gathering and testing were automated and fed into the approval gates the bank already runs. The committee kept its place: "Review and approval stay with the risk committee, but the committee now bases each decision on consistent evidence, collected the same way across use cases." What changed is what lands in front of it. The post lists an executive presentation, a mitigation checklist, a detailed technical report, and a dashboard, with reviewers confirming the findings before a gate decision. Standard Bank Group reports that "the assessment reduced a traditionally manual two-week review to a comprehensive risk report generated in approximately two hours, demonstrated across six assessments." It also reports "a 14% improvement in use-case risk scores after remediation," which describes scores after teams fixed the gaps the tool flagged, not a measured business result. Those figures are company-reported in a vendor co-written post and have not been independently verified here.

For the AI Value Index, this is an international comparable, not a U.S. national, regional, or local record. The useful claim is about the gate. Seen from the committee's side of the table, the automation takes the wait for the file and leaves the vote where it was. The report arrives in about two hours, the post says, built on evidence gathered the same way each time, so the committee can start the hard conversation on day one. That is a productivity story about waiting time.

The bank still owns the call. The post says the assessment "produces evidence against controls" and "doesn't certify compliance with any regulation." Reassessment after launch is still ahead: once live workloads are connected, the assessment "will rerun." What would move this beyond a signal is a verified change in approval time, on a stated baseline, at a named institution, with the human gate still holding.