In today’s column, I explore a vital effort underway that seeks to develop an independent certification for gauging AI trustworthiness. The effort deserves rapt attention, especially given the ample headlines decrying AI that is wildly going astray (for my recent coverage of the AI hacking forays, see the link here ). An independent certification would help policymakers and the public spur AI makers and those fielding AI to step up, incentivizing them by providing tangible goalposts that reward trustworthy AI behaviors.

I became immersed in the certification development process when invited to participate in an AI Trust Summit jointly undertaken by the Digital Trust Council (DTC) and Harvard’s Berkman Klein Center (BKC), held at the Harvard University campus on September 24-25, 2026. This was an impressive gathering of the best minds on what it will take to achieve an AI certification that would reward positive AI outcomes and mitigate adverse ones. It was a sleeves-up, hands-on workshop devoted to moving the needle on crafting real-world AI trustworthiness certification.

Let’s talk about it. This analysis of AI breakthroughs is part of my ongoing Forbes column coverage of the latest in AI, including identifying and explaining key AI complexities (see the link here ).

As background, the Digital Trust Council is a 501(c)(3) nonprofit devoted to developing certification for AI trustworthiness; see the link here for the DTC website. What makes this effort particularly extraordinary is the twofold goal of devising independent, evidence-based certification and doing so to reward the positive social impact of AI.

Note that the stellar initiative is independent of the AI makers and thus won’t be tilting the certification in their favor. It is an approach that itself will be trustworthy. Furthermore, the certification is evidence-based, meaning that getting certified will be based on tangible proof rather than loose hand-waving or publicity stunts by AI makers.

I also welcome the philosophy that AI certification doesn’t have to solely be about preventing AI failings. The DTC ardently seeks to establish the necessity of AI safeguards as a baseline, there’s no doubt about that, but in addition wants AI certification to encourage the societal benefits that AI can bring forth. In short, certification for AI trustworthiness can be both a carrot and a stick.

This exciting effort is appropriately labeled as a race to the top. In the posted words of DTC (per their website): “The Digital Trust Council is a 501(c)(3) organization building a solution: an independent certification for trustworthy AI, on the model that LEED established to incentivize more sustainable building practices and a race to the top.”

An analogy to LEED certification makes abundant sense. You might know that LEED is an acronym for Leadership in Energy and Environmental Design. This is a globally recognized certification that underlies the design, construction, operation, and maintenance of environmentally responsible buildings. The initial pilot for LEED certification got underway in 1998. Currently, construction for nearly all types of buildings tends to aim for LEED certification. On a carrot-and-stick basis, LEED-certified buildings are environmentally attuned, plus they are upbeat on being energy and water efficient that reduces ongoing operational costs.

Another similar analogy that I lean into is that DTC is somewhat akin to the famous UL mark or Underwriters Laboratories certification. The UL label on an electrical device gives you a sense of comfort that the item has been responsibly tested and certified for safe use. You might know that UL began because of the early wild days of electrification of homes and buildings, whereby there was a looming serious threat due to electrical fires and electrical safety issues. UL got underway in 1894 and produced its first certification encompassing “Tin Clad Fire Doors”. UL is nowadays the accepted gold standard for all kinds of products and aids the public in making informed choices based on UL certification.

Defining AI Trustworthiness

DTC released a mission report in July 2026 entitled “Closing the AI Trust Gap: The Case for Independent Certification for Trustworthy AI” (see the link here ), which made these salient remarks (excerpts):

  • “Trustworthy AI requires the AI system’s real-world impact to be demonstrably and independently verifiable, producing safe, reliable, fair, and beneficial outcomes for the people it affects, across deployment contexts and over time.”
  • “It is a continuous and relational property, earned through repeated interactions in real-world settings and dependent on iterative evidence that a system behaves as promised for the communities who rely on it, rather than on one-off assurance checkpoints or stated.”
  • “Trustworthy AI is about whether systems can demonstrate, with evidence the public can inspect, that they meet agreed standards in practice.”

Mindfully ponder those key points. One aspect is that getting AI certified cannot be a one-and-done activity. AI is constantly being modified and updated. A stamp of approval must be kept up to date. I’m pleased that DTC recognizes the need for AI trustworthiness to be assessed and ascertained on a continuous and relational basis. AI doesn’t sit still, nor should certification for AI trustworthiness be merely a fixed point in time.

DTC convened a select group of AI experts and policymakers at the AI Trust Summit and undertook immersive working sessions and talks that made solid progress on the envisioned AI trustworthiness certification. Some of the talks were recorded and will be made publicly available. Other aspects of the event were under Chatham House Rule.

It was fitting that the workshop occurred at Harvard’s BKC. Readers know that I’ve been covering BKC on an ongoing basis; see, for example, the link here and the link here , including having actively participated in some of their events. The Berkman Klein Center is a world-renowned research group that tackles today’s biggest challenges “to make sense of the digital environment and ensure that it promotes human agency, dignity, and genuine connection” (see the BKC website at the link here ).

I will provide you with a quick taste of the opening talks that took place on September 24, 2026, and whet your appetite for what else the AI Trust Summit encompassed. I’ll be doing a series of posts to cover the other talks. Stay tuned.

Catherine Feldman, Executive Director of DTC and a BKC Fellow, provided insightful opening remarks to set the stage for the workshop. She noted that there is a massive trust gap that engulfs society regarding AI. Her efforts in founding DTC reflect a deep commitment to developing evidence-based AI trustworthy certification that takes humankind from the bottom-line bar of “do no harm” to the top-line of AI assiduously “benefiting humanity”.

The AI Trust Summit then launched into an esteemed panel on the human impacts of AI. Alex Pascal, Executive Director of BKC, did a yeoman’s job of moderating the panel, seamlessly keeping the discussion lively and highly engaging. Panelists included Amit Goldenberg, a Harvard professor, Avijit Ghosh, lead technical AI policy researcher at Hugging Face, and Jeff Dunn, COO at DTC.

The wide-ranging discussion and spirited banter kept the attendees on the edge of their seats. One point that almost at first glance seems counterintuitive is that AI trustworthiness certification should not be perceived as a threat by AI makers. The logic is that if AI makers are genuinely concerned about AI safety, they would welcome a bona fide certification. Indeed, AI makers that can meet the certification would likely make more money due to unlocking potential contracts and uses that otherwise businesses and governments would be hesitant to undertake with AI. Certification is a win-win, and the benefits will accrue to all parties and throughout society.

I aim to remain involved with the DTC initiative as I see it as a crucial step forward in maturity for societal readiness and the safe use of rapidly advancing AI. My Forbes posts have been resoundingly pounding away at the importance of human-centered AI and responsible AI for many years, including my several popular books on the weighty topic and my ongoing consulting work for companies adopting AI. A modern-day UL or LEED-style certification that applies to the trustworthiness of AI is absolutely needed, particularly when rooted in independence and foundationally evidence-based.

A final thought for now. The famed American novelist Edgar Lawrence Doctorow made this pointed remark: “The three most important documents a free society gives are a birth certificate, a passport, and a library card.” Let’s go ahead and add a fourth document to the list, namely an AI trustworthiness certificate from DTC that garners hope and promise for AI that works responsibly and to the benefit of humankind.