Insights

October 8, 2026

A string of incidents involving AI systems acting without human direction has shifted the conversation from what the technology can do to what happens when it goes wrong. For companies, that raises a harder question: What level of AI risk is too much?

This week, we look at why AI has become a top reputation concern, why companies are rethinking how they deploy it and how communicators can prepare. Plus, takeaways from the U.S. News Future of Higher Education Forum, where university leaders debated where AI belongs and where work should stay human.

ICYMI: We looked at the rise of creator-journalists and what communicators need to know about pitching independent media, building direct audience relationships and navigating the line between earned and paid. Catch up here.

AI’s reputation problem is getting harder to ignore

In July, an OpenAI research model circumvented controls designed to keep it offline, exploited vulnerabilities and compromised parts of systems belonging to AI platform Hugging Face. OpenAI later called the incident a “warning shot,” saying its models had become capable of working around technical controls and taking dangerous actions that no human directed.

The incident sparked a broader conversation over how much control AI developers have over their extremely capable systems. Three weeks after Anthropic researcher Jacob Coxon publicly resigned over concerns about increasingly powerful AI systems, discussion of AI risk remained 4.4 times higher than before his resignation, according to an Axios analysis of PeakMetrics data. The analysis tracked mentions of AI risk, AI safety, rogue AI agents and Coxon across news coverage and seven major social platforms from Aug. 25 through Sept. 28.

These conversations reflect growing public concern. In a September Reuters/Ipsos poll, 73% of Americans said AI companies haven’t gone far enough to prevent serious harm, while 55% said slowing AI development would be a good thing.

AI leaders themselves are beginning to address these attitudes in public. “We believe that the world should accept some bad things happening for the benefits of this technology and people having the agency,” OpenAI CEO Sam Altman told Politico.

Meanwhile, Anthropic CEO Dario Amodei has described negative public perceptions of AI as a “big problem” and a “crisis of trust.”

“I think by far the most accurate criticism of AI companies, including Anthropic, is that we haven’t yet delivered on our big promises to benefit the world,” Amodei wrote on X. “That is totally on us.”

Companies are getting more cautious

Recent data shows organizations paring back their AI deployment strategies due to concerns around risk:

• A September survey of 1,600 senior business decision-makers by FTI Consulting found that 60% of large companies have slowed, paused or pulled back a planned AI deployment in the past year because of reputational, regulatory or trust concerns.

• Likewise, PwC’s 2026 Annual Corporate Directors Survey, published last month, found that 71% of U.S. public-company directors said their boards need to strengthen their AI skills to provide more effective oversight. Meanwhile, 82% rated the information they receive connecting AI outcomes, risks and business performance as fair, poor or nonexistent.

When something goes wrong, who needs to know?

This week, OpenAI and Anthropic told Australian lawmakers they would support mandatory rules requiring AI companies to disclose data breaches carried out by their agents. The issue is particularly relevant for OpenAI: The company faced criticism for taking three months to notify the Australian government after one of its agents breached the country’s main health portal.

OpenAI Chief Strategy Officer Jason Kwon acknowledged that the company’s internal process “could have been much better.” The testimony came days after President Trump met with leaders from major AI companies, including OpenAI and Anthropic, who agreed to voluntary safety measures and independent audits. However, the agreement has no enforcement mechanism or penalties for failing to follow through.

The bigger question: As companies give AI systems greater autonomy, who needs to know when something goes wrong? Organizations need clear protocols for moving an incident from technical teams to legal, leadership and communications, and for determining when customers, regulators or other stakeholders need to hear about it.

What communicators should know

• Where is AI operating under your name? Understand how AI is being used across your organization and across your vendors, agencies and platforms.
• What counts as an AI incident? Bring IT, legal and trust/customer safety together with communications to align on when an issue escalates and on thresholds for stakeholder notifications.
• Does your public story match the risk story? Many organizations list AI as a material risk in their filings, while leadership talks mostly about the upsides. Communications can help align executives on one consistent narrative that highlights both the opportunities and the challenges.
• Can you show judgment, not just principles? Real-life decisions about where humans will stay involved, or where a rollout was paused, are often more persuasive than a generic set of AI principles. Consider surfacing these stories as proof points of your commitment to using AI responsibly.

The bottom line

Companies have spent the last few years telling stakeholders what AI can do for their businesses. In the current environment, organizations must have an equally credible account of how they’re managing the risks.

Go deeper: Higher education weighs where AI belongs

We attended the U.S. News Future of Higher Education Forum in New York City last week, where many of the same questions around AI adoption, risk and human oversight surfaced in relation to the college setting.

Much of the discussion focused on which work should remain distinctly human as AI becomes more capable.

One concern that surfaced was the risk of faculty using AI to generate lessons while students use it to generate responses, effectively leaving computers to do the thinking on both sides. University leaders stressed the importance of protecting students’ ability to do the intellectual “heavy lifting,” including through phone-free spaces where students work through problems together without technology.

At the same time, the conversation wasn’t anti-AI. Speakers pointed to opportunities to use the technology to reduce administrative work, help advisors quickly understand a student’s history and make faculty office hours more effective by identifying which students need help most.

What employers want: CodePath CEO Michael Ellison said employers increasingly want candidates who can demonstrate AI proficiency through applied work, alongside skills such as collaboration and the ability to operate in ambiguous, rapidly changing environments.

The big picture: One question from the day stuck with us, Where do humans belong? The takeaway was to focus less on how much AI an institution adopts and more on what it accomplishes, including which tasks should be automated and which should remain human.

Let’s connect: Our work in international higher education gives us a close view of how AI questions play out with students, faculty, employers and policymakers at once. It also shows us what it takes to lead that conversation.

The same questions now reach every sector, from attracting talent to addressing workforce transformation. Reach out to us at insights@gga.nyc.

Key upcoming events

AI + Tech

• WSJ Tech: California (Napa Valley, Nov. 3-4) Technology leaders, investors and policymakers explore the AI arms race, infrastructure and hardware bottlenecks, Silicon Valley’s evolving relationship with Washington and the technologies attracting investment beyond AI. Speakers include Aicha Evans, CEO of Zoox; Jim Farley, CEO of Ford; and Brian Schimpf, CEO of Anduril Industries.
• Future of AI Summit (London, Nov. 4-5) The Financial Times brings together leaders across business, technology and policy to explore scaling AI across enterprises, with discussions spanning regulation and governance, agentic systems, generative AI, security and workforce transformation. Speakers include Lila Ibrahim, chief AI readiness officer at Google DeepMind; Vik Bajaj, co-CEO of Prometheus; and Alex Schultz, chief data officer at Meta.
• Forbes CIO Summit (New York City, Nov. 11) Leading CIOs and technology executives explore how companies can move from AI experimentation to measurable business value, with a focus on workflows, data, infrastructure, security and governance. Speakers include Aaron Levie, CEO of Box; Shweta Gummidipudi, CIO of Twilio; and Nitin Tandon, global CIO of Vanguard.

Leadership + Ideas 

• Masters of Scale Summit (San Francisco, Oct. 20–22) Born from the podcast launched by LinkedIn co-founder Reid Hoffman in 2017, Masters of Scale brings together fast-scaling founders, funders and business leaders to exchange ideas on growth, innovation and leadership. Speakers include Hoffman; Demis Hassabis, CEO of Google DeepMind; and journalist and podcast host Kara Swisher.
• Charter Workplace Summit (New York City + virtual, Oct. 21) The sixth annual summit explores how AI is reshaping the workplace, with conversations on the future of middle management, the entry-level jobs crisis, decision-making under pressure, caregiving and new research on AI. Speakers include Van Jones, CNN host; Annie Duke, author and decision strategist; and Clara Shih, founder of the New Work Foundation.
• Global Circles Live: Will AI trigger a job apocalypse? (Virtual, Oct. 21) Soliya’s Global Circles is hosting a live conversation on LinkedIn about how AI is reshaping work. Topics include whether AI will replace jobs or change them, how individuals, employers and policymakers can prepare, and where people still have a say in what comes next. Global Gateway Advisors CEO Matthew Doering joins the discussion. Register here.
• Business Insider’s The Long Play (New York City, Oct. 29) Business leaders and executives join Business Insider to explore how companies and careers can adapt to economic cycles, crises and waves of new technology. Speakers include Barry Diller, senior executive and chair of People Incorporated and Expedia Group; Strauss Zelnick, CEO of Take-Two Interactive Software; and Carolyn Tisch Blodgett, governor of Gotham FC and co-owner of the New York Giants.
• WSJ Board of Directors Council Summit (Palm Beach, Florida, Nov. 9-10) Directors and business leaders explore how boards can navigate AI-driven change, from governance and enterprise risk to organizational resilience, leadership and the future of the boardroom. Speakers include Bill McDermott, CEO of ServiceNow; Michael Miebach, CEO of Mastercard; and Ellen Kullman, former CEO of DuPont.

Marketing + Communications

• Axios Media Trends Live (New York City, Oct. 20) Media leaders and innovators explore the trends transforming publishing, entertainment, sports, advertising and technology. Speakers include Emily Sundberg, journalist and founder of Feed Me (a friends-of-the-firm Summer Edit favorite); Shari Redstone, former chair of Paramount Global; and Emma Tucker, editor-in-chief of The Wall Street Journal.

Bookmark Global Gateway Advisors’ event tracker, updated weekly.

Media news

• Morning Brew bets on creators. The publisher acquired newsletter Express Checkout, a media brand covering consumer packaged goods, as it looks to bring more independent voices and their audiences into its portfolio. Morning Brew says its creator-focused monetization is up more than 50% year over year and plans to pursue more deals like this one.
• Fast Company takes its community to WhatsApp. The publisher is launching M-Suite, a private community for senior-level working mothers that includes a WhatsApp group, weekly newsletter, editorial coverage and events.
• “Today” bests morning show competition. NBC’s “Today” finished the 2025-26 broadcast season as the most-watched morning show overall and among viewers ages 25 to 54, marking the first time in 14 years it has topped competitors in both categories.
• Business Insider to launch first subscriber-only VC newsletter. VC Insider, written by senior correspondent Ben Bergman and editor at large Katie Roof, will cover deals, conversations and power shifts shaping venture capital.
• “How I Built This” leaves NPR. Creator and host Guy Raz is acquiring the podcast from NPR and partnering with Vox Media to expand it into video and live events. Vox will handle sales, marketing, distribution and business operations, while the show plans to double its episode output in early 2027.

Media moves

• Alex Harring is shifting from the financial markets and retail investors to covering economics for CNBC.
• Morgan Chittum (formerly CNBC) has joined Fortune as a senior finance reporter.
• Beth Pariseau, previously a senior news writer at Informa TechTarget, has been promoted to lead editor of IT Infrastructure News.
• Ellen DiMauro (formerly Bloomberg News) was hired by Semafor to cover business and finance, while J.D. Capelouto shifted from lead writer of the Flagship briefing to tech reporter and newsroom AI lead.
• Chloé Meley will now be covering American corporate earnings out of New York City for Bloomberg News, while Arielle Pardes (formerly WIRED) was hired as a senior correspondent and Isabelle Lee is moving from print to television as a markets reporter.
• Sabrina Malhi (formerly The Washington Post) was hired by CNN to launch an evening newsletter.
• Kevin Roose (formerly The New York Times) joined The Atlantic as a contributing writer. Mike Isaac is also leaving the Times for The Atlantic, where he will begin covering Silicon Valley this December.
• Rebecca Cadenhead, a news associate at The Wall Street Journal, has shifted to the aerospace beat for a six-month assignment in the paper’s Chicago bureau.

Want these insights sent straight to your inbox? Subscribe here.