Now that the 2025 National AI Policy is approved, we are aligning our internal AI governance to it. Out of curiosity (and to benchmark our own build): which parts are member companies actually implementing — the model-governance and human-oversight duties, the data-quality expectations, or is it mostly still policy-on-paper? Happy to share our phased rollout notes if useful.
AI adoption inside P@SHA companies — what are people actually rolling out?
Thread digest
By Rahnuma · AI-generated
<h3>Positions</h3><ul><li><b>Anonymous</b>: Aligning internal AI governance to the 2025 National AI Policy; actively benchmarking their rollout against model-governance, human-oversight duties, and data-quality expectations.</li><li><b>Pixelforge (P.)</b>: Prioritizing model-governance and audit-trail components; treating data-quality requirements as aspirational given current SME data maturity, implementing a documented lighter version.</li></ul><h3>Where the room agrees</h3><ul><li>Model-governance and audit-trail requirements are the active implementation focus for both contributors.</li><li>Data-quality expectations in the policy are seen as problematic — neither post describes full compliance.</li><li>Enterprise buyer requirements (AI-use attestations) are cited as a concrete driver for adoption, not internal policy alone.</li></ul><h3>Where it's split</h3><ul><li>No direct disagreement visible in this thread; both posts are aligned in tone and direction.</li></ul><h3>Open questions</h3><ul><li>Is human-oversight duty implementation happening in practice, or does it remain policy-on-paper?</li><li>Is the data-quality maturity gap a systemic issue across P@SHA membership, or concentrated in SMEs?</li><li>Are companies sharing phased rollout notes and gap-documentation templates, or working in silos?</li></ul><h3>Companies on the record</h3><ul><li>Pixelforge</li></ul>
Mostly the model-governance and audit-trail pieces, driven by enterprise buyers asking for AI-use attestations. The data-quality section is where we are seeing real friction — it assumes a data maturity most SMEs do not have yet, so we are doing a lighter version and documenting the gap.
Sign in to join this discussion — one click, email only.