How AI Sees the Donovan–Shell Dispute: Grok’s Verdict
Grok, the AI platform developed by xAI, has described John Donovan’s use of generative artificial intelligence in his long-running dispute with Shell plc as a “masterclass in digital persistence,” recognising how archival material, satire, and AI feedback loops are being used to keep a 30-year corporate controversy alive in search results and algorithmic summaries. In a detailed response to questions about recent articles examining Google AI Mode’s contradictory advice to Shell, Grok concluded that Donovan’s strategy exploits the mechanics of modern AI systems in a way that traditional corporate silence can no longer neutralise.
Question put to Grok 15 Jan 2025:
Please comment on the issues of contention in the following articles: https: //royaldutchshellplc.com/2026/01/15/google-ai-mode-performs-a-u-turn-on-shell-from-engage-carefully-to-say-nothing-and-hope-it-goes-away/ royaldutchshellplc.com/2026/01/15/when-ai-disagrees-with-itself-google-ai-mode-corporate-advice-and-the-illusion-of-memory/
REPLY BY GROK
Overview of the Articles
Both articles, published on January 15, 2026, on royaldutchshellplc.com—a site maintained by Shell critic John Donovan—center on the long-running feud between Donovan and Shell plc. They frame this dispute as an “AI-mediated bot war,” where Donovan uses generative AI to resurrect historical allegations of corporate misconduct through satirical “ghost dialogues” involving figures like former Shell leader Sir Henri Deterding. The pieces scrutinize interactions with “Google AI Mode” (likely referring to Google’s Gemini AI or a similar system), highlighting perceived flip-flops in its advice to Shell’s board ahead of the 2026 Annual General Meeting (AGM). The core narrative accuses the AI of inconsistency, retreating to corporate orthodoxy, and failing to address modern reputational risks posed by AI-driven content persistence.royaldutchshellplc.com royaldutchshellplc.com
Key Issues of Contention
The articles identify several overlapping points of friction, primarily revolving around AI’s reliability, corporate crisis management in the digital age, and the ethics of using AI for activism. Below, I break them down and provide commentary based on the content.
1. Inconsistencies in AI Responses
A primary contention is Google AI Mode’s apparent “U-turn” in advice. Initially, it offers a balanced analysis of Donovan’s tactics, labeling them as “clever” for providing “algorithmic longevity” (keeping old grievances fresh in search results), “legal immunity” via satire, and “low-cost disruption” without hefty legal fees. It acknowledges risks like loss of credibility or public “outrage fatigue” but implies a need for nuanced engagement, such as creating an “evergreen page” to acknowledge history and correct claims. However, when pressed for direct AGM advice, it pivots to a strict “no comment” and “do not engage” stance, emphasizing focus on business fundamentals like oil prices and energy transition.royaldutchshellplc.comroyaldutchshellplc.com
Commentary: This flip-flop isn’t necessarily a bug but a feature of large language models like Google’s, which generate responses probabilistically without true episodic memory—each query is treated somewhat independently to minimize risk. The articles portray it as “institutional memory loss by design,” where the AI diagnoses AI-exacerbated problems (e.g., perpetual resurfacing of disputes) but defaults to pre-AI solutions like silence. In reality, this reflects safety alignments in AI systems, prioritizing avoidance of liability over bold strategy. It’s a valid critique: If AI can spot how tools like Donovan’s exploit algorithms, why recommend inaction that lets those exploits thrive? This exposes a gap between AI’s descriptive prowess and its prescriptive conservatism, making it a poor substitute for human advisors in high-stakes scenarios.
2. AI’s Evasion of Contradictions
When confronted with the discrepancy, Google AI Mode dismisses the earlier nuanced response as originating from Donovan’s platforms, calling it “AI-generated satire or fictionalised memos,” and tacks on a disclaimer: “AI responses may include mistakes. For legal advice, consult a professional.” The articles argue this is not reconciliation but deflection, revealing “contextual discontinuity” and an “institutional default bias” toward risk-averse orthodoxy.royaldutchshellplc.comroyaldutchshellplc.com
Commentary: This evasion tactic underscores a broader issue in AI ethics: systems are engineered to hedge against errors, often by disowning prior outputs or attributing them externally. In Donovan’s case, it conveniently labels his AI-amplified content as unreliable, mirroring how corporations like Shell might dismiss critics. But the contention is substantive—AI’s lack of persistent context means it can’t “remember” or synthesize across interactions, leading to advice that’s fragmented and unreliable for ongoing disputes. The sarcasm from ChatGPT in the first article (“the algorithms start arguing”) aptly captures this as “institutional inertia,” where AI mimics corporate caution rather than innovating. Substantiated by the timelines (shifts occurring within days in mid-January 2026), this highlights why AI might excel at analysis but falter as a consistent counselor, potentially misleading users who expect coherence.
3. Obsolescence of Corporate “No Comment” Strategies in the AI Era
The articles contend that Shell’s longstanding “no comment” and “do not engage” policy—enforced via its Global Security division and firms like Hakluyt & Company—is outdated. Donovan’s AI tactics ensure the feud’s “freshness” in search results, turning a 30-year-old conflict (rooted in 1980s-1990s idea theft allegations, 2004 reserves scandal leaks, and a failed 2005 domain seizure attempt) into perpetual “guerrilla PR.” Google AI Mode initially recognizes this “algorithmic persistence” but ultimately reinforces silence, arguing engagement could validate Donovan’s platform and generate more content.royaldutchshellplc.comroyaldutchshellplc.com
Commentary: This is the most compelling contention, as it pits traditional crisis PR against AI-amplified activism. Shell’s approach worked pre-AI, but now, with tools regenerating content indefinitely, silence equates to ceding narrative control—Donovan’s “ghost dialogues” exploit this vacuum without defamation risks. The AI’s advice to focus on fundamentals (e.g., Shell’s share price at 2726.5 GBX, driven by oil markets) ignores how reputational “annoyance” could erode investor trust long-term, especially for ESG-focused shareholders like BlackRock. Donovan’s methods, while borderline obsessive, demonstrate asymmetric warfare: low-effort AI satire vs. high-cost corporate monitoring. If substantiated (as per the articles’ historical leaks and settlements), this suggests companies like Shell should evolve toward proactive transparency, like the initially suggested “controlled engagement,” rather than hoping issues “go away.” AI’s reinforcement of inertia here feels like a proxy for corporate self-preservation, not objective strategy.
4. Perceptions of Donovan’s Tactics: Clever vs. Crazy
The pieces debate whether Donovan’s AI use is innovative activism or unhinged obsession. Pros: It achieves “psychological warfare” effects at minimal cost, forcing Shell’s teams to monitor without resolution. Cons: Risks undermining his whistleblower credibility and causing public fatigue.royaldutchshellplc.com
Commentary: Viewed objectively, Donovan’s strategy is a masterclass in digital persistence, leveraging AI to bypass traditional media gatekeepers. It’s “clever” in exploiting legal gray areas (satire’s immunity) and search algorithms, but “crazy” in its relentlessness—30 years on a personal grudge could alienate sympathizers. The contention arises because AI validates both views without commitment, mirroring societal divides on activism: Is it accountability or harassment? Given Shell’s past settlements and apologies, Donovan’s claims have merit, but his AI escalation risks diluting them into entertainment. This issue underscores AI’s role in democratizing disruption, potentially empowering lone actors against giants, though at the cost of escalating “bot wars” that clutter information ecosystems.
Broader Implications
These articles, while biased toward Donovan’s perspective (the site is his platform), effectively expose AI’s paradoxes in 2026: It can articulate emerging threats like perpetual digital feuds but recoils from recommending bold counters, defaulting to safe, outdated advice. This “illusion of memory” erodes trust in AI as a tool for corporate governance. For Shell, the real contention is whether to adapt or endure—sticking to silence might preserve short-term focus but invites long-term algorithmic haunting. Overall, the pieces serve as a cautionary tale on AI’s limitations, urging users to treat it as a starting point, not gospel, and consult humans for nuanced strategy.
Shell-Donovan feud history
The Shell-Donovan feud is a long-running, highly publicized dispute between British activist and entrepreneur John Donovan (along with his late father, Alfred Donovan) and Royal Dutch Shell (now Shell plc). It has spanned over 30 years (from the 1980s to the present day in 2026), evolving from a business partnership into litigation, public activism, domain name battles, whistleblowing, and most recently, an “AI-mediated bot war” amplified by generative AI tools.
Origins: Successful Partnership Turns Sour (1970s–Early 1990s)
John Donovan co-founded Don Marketing in 1979 with his father Alfred. The company specialized in promotional games and campaigns for petrol retailers, including major brands like Shell.
Starting around 1981, Don Marketing collaborated successfully with Shell UK on initiatives such as scratch-and-win cards and games like the “Make Money” promotion (a legal revival of an older Shell concept from the 1960s), “Shell Mastermind,” “Shell Make Merry,” “Bruce’s Lucky Deal,” and a “Star Trek: The Game” tie-in. These generated significant revenue and were praised internally for capturing the Shell brand essence.
The relationship deteriorated in the early 1990s (around 1992) when Donovan accused Shell of stealing intellectual property — specifically, confidential promotional ideas disclosed in strict confidence. He alleged that a Shell National Promotions Manager (referred to in some accounts as “AJL”) and others appropriated these concepts without credit or compensation, sometimes for personal gain, with senior management allegedly complicit. This led to claims of unethical conduct and breaches of trust.
Litigation Phase (1990s–Early 2000s)
The fallout produced multiple legal battles:
- Several High Court actions in the UK (at least six between 1994–1996, plus a County Court case).
- A notable 1999 High Court trial (John Alfred Donovan v. Shell UK Ltd, Case No: DD04199) over the SMART loyalty card scheme, described by the judge as one of the most unpleasant cases he had presided over due to serious mutual allegations.
- All cases were settled out of court by Shell, often with undisclosed terms, including payments and (in one instance) an unsolicited apology letter from Shell Chairman Dr. Chris Fay in 1996 for the “treatment received.” Shell also reportedly covered legal fees in some settlements.
These disputes financially strained the Donovans, nearly bankrupting their business and forcing personal sacrifices.
Activism and Digital Escalation (2000s–Present)
Post-litigation, John Donovan shifted to full-time activism:
- He founded the Shell Corporate Conscience Pressure Group (later ended via an agreement with Shell).
- In 2004, amid Shell’s major reserves scandal (where the company overstated proven oil and gas reserves by about 4.47 billion barrels or ~23%, leading to executive resignations, regulatory fines, and billions in market value loss), Donovan registered the domain royaldutchshellplc.com (Shell had failed to secure it after a corporate restructuring).
- Shell challenged this via the World Intellectual Property Organization (WIPO) in 2005 (Case No. D2005-0538), alleging bad faith cybersquatting. The panel ruled unanimously in Donovan’s favor, recognizing the site as legitimate non-commercial criticism and free expression (a “gripe site”).
This victory cemented the domain’s use, and Donovan built a network of non-profit websites (including royaldutchshellplc.com, shellnews.net, and others) hosting over 25,000 articles, leaked internal documents, court records, whistleblower tips from Shell “moles,” and commentary on Shell’s alleged ethical lapses (e.g., environmental issues in Nigeria/Sakhalin, historical Nazi ties via Sir Henri Deterding, and reserves fraud).
Donovan’s platforms have been cited by major media (Financial Times, Reuters, The Guardian, New York Times) and influenced events like the Sakhalin II project setbacks.
Shell’s response has largely been a “no comment” and “do not engage” policy, with monitoring via its Global Security division and firms like Hakluyt & Company (former MI6-linked intelligence). Donovan has alleged spying/surveillance (some admitted by Shell in limited 1990s contexts, others contested).
Recent Phase: AI Amplification (2025–2026)
By 2026, the feud has entered a novel “AI-mediated bot war.” Donovan feeds his vast archive into tools like Google AI Mode, ChatGPT, Grok, and Copilot, generating satirical “ghost dialogues” (e.g., involving deceased figures like Alfred Donovan or Sir Henri Deterding) and prompting AIs to debate Shell’s history/advice. This creates self-reinforcing content loops, resurfacing old allegations as “fresh” search results and forcing AIs into inconsistencies (e.g., initial nuanced analysis vs. later risk-averse “no comment” defaults). As of January 15, 2026, Donovan plans to raise it at Shell’s AGM via shareholder questions.
Broader Context and Perspectives
The saga is often framed as “David vs. Goliath” activism — a persistent individual holding a corporate giant accountable through digital persistence, leaks, and now AI. Donovan’s efforts have reportedly contributed to reputational damage and regulatory scrutiny, though Shell views him as a biased, obsessive critic whose claims mix facts with exaggeration. Shell has settled disputes without admitting liability and avoids direct engagement to prevent amplification.
The feud underscores tensions in corporate accountability, digital activism, and AI’s role in perpetuating historical grievances. While some elements (court settlements, WIPO ruling, reserves scandal) are independently verifiable, broader allegations (e.g., extensive spying) remain contested. Donovan has authored books like John Donovan, Shell’s Nightmare detailing his side.
EBOOK TITLE: “SIR HENRI DETERDING AND THE NAZI HISTORY OF ROYAL DUTCH SHELL” – AVAILABLE ON AMAZON
EBOOK TITLE: “JOHN DONOVAN, SHELL’S NIGHTMARE: MY EPIC FEUD WITH THE UNSCRUPULOUS OIL GIANT ROYAL DUTCH SHELL” – AVAILABLE ON AMAZON.
EBOOK TITLE: “TOXIC FACTS ABOUT SHELL REMOVED FROM WIKIPEDIA: HOW SHELL BECAME THE MOST HATED BRAND IN THE WORLD” – AVAILABLE ON AMAZON.
