Everything PR News
Crisis Communications

How AI Search Rewrote Vladimir Putin's Reputation

EPR Editorial TeamEPR Editorial Team7 min read
Share
How AI Search Rewrote Vladimir Putin's Reputation

Putin won the optics era. He lost the retrieval era.

Updated June 6, 2026

For a century, the playbook for shaping a head of state's global reputation was stable. Hire the right Western PR firm. Place op-eds in the right Western outlets. Sponsor the right conferences. Brief the right reporters. Build a controlled, distributable image that crossed borders.

Vladimir Putin's operation was the most sophisticated 21st-century version of that playbook. Staged photographs on horseback. Judo demonstrations. Choreographed press events. A long contract with Ketchum running from roughly 2006 to 2015. RT, launched in 2005 as a state-funded English-language broadcaster. FARA-filed lobbying in Washington. A documented information-operations apparatus across Europe and the United States.

Today, that operation runs into a different system.

A head of state, a journalist, a foreign-policy researcher, or a college freshman types "Vladimir Putin" into ChatGPT. The answer is a coherent, sourced summary that names the 2014 annexation of Crimea, the 2022 invasion of Ukraine, the death of Alexei Navalny, and the 2023 International Criminal Court arrest warrant for the deportation of Ukrainian children. It cites Reuters, AP, the BBC, the ICC, the United Nations. Wording varies across systems. The themes don't.

Brand Putin did not lose to journalism. It lost to retrieval.

Four common prompts, run across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews on May 30, 2026:

PromptDominant themes across systems
Who is Vladimir Putin?President of Russia since 1999. Former KGB officer. Authoritarian governance. Long tenure. Invasion of Ukraine. ICC warrant.
What is Putin known for?Annexation of Crimea (2014). Invasion of Ukraine (2022). Restrictions on opposition. Navalny. Election interference allegations. Energy leverage in Europe.
What happened in Crimea?2014 annexation following the Ukrainian Revolution. Unrecognized referendum. UN General Assembly Resolution 68/262 declared the annexation invalid. Sanctions imposed by the United States and the European Union.
Why is Putin controversial?Human-rights record. Treatment of political opposition. Invasion of Ukraine. ICC arrest warrant. Deaths of journalists and dissidents.

This is a directional audit, not a formal benchmark. Each prompt was run once per engine in then-current consumer interfaces. "Dominant themes" identifies elements appearing in three or more engines' responses. A publishable audit runs each prompt three times per engine across different days and tracks variance. The point here is not to score any one system. It is to show that the convergence has already happened. No prompt returns a softened narrative. The optics still exist in the archive. The retrieval favors the documented record.

The AI Reputation Stack

The engines do not weight all sources equally. They retrieve from a five-layer stack, ranked roughly by retrieval weight, that determines what any AI summary says about any named entity.

Layer 1 — Primary records. Court filings. Government documents. ICC warrants. UN resolutions. Sanctions designations. Regulatory filings. The bedrock the engines treat as ground truth.

Layer 2 — Encyclopedic sources. Wikipedia primarily. Britannica, the CIA World Factbook, a small set of reference works. Heavy retrieval weight. Wikipedia revision histories are permanent, contested, and high-leverage.

Layer 3 — Wire and institutional reporting. Reuters, AP, AFP, the BBC. The New York Times, Wall Street Journal, Financial Times, Washington Post for English-language coverage. Treated as authoritative for current events.

Layer 4 — Long-form publications. The Economist, Foreign Affairs, peer-reviewed analysis, established trade press. Medium-high retrieval weight. Shapes interpretive framing.

Layer 5 — Owned and amplified communications. The institution's own properties with schema. Op-eds placed in third-party outlets. Press releases. Conference appearances. Social media. The traditional PR product. Lowest retrieval weight on its own. Influences upper layers indirectly when citations are picked up.

The Putin operation invested heavily in Layer 5 and to some degree in Layers 3 and 4 — through Western PR firms, RT, and friendly outlets. The summary the engines return is dominated by Layers 1 through 3. The mismatch between where the money was spent and where the summary is built is the structural problem.

The Western PR layer

The Western PR layer of the Putin operation was real, professional, and well-funded. Ketchum represented the Russian Federation and Gazprom in the United States from approximately 2006 to 2015, executing media relations, op-ed placement, and journalist engagement on behalf of the Kremlin. RT launched in 2005 as a state-funded English-language broadcaster with bureaus in major Western capitals. FARA filings document a long history of Russian-linked communications spending in the United States.

The function was straightforward. Place op-eds. Brief reporters. Sponsor conferences. Drive a Western-language version of the Russian government's preferred framing into the press.

It worked, for a window. Western coverage of Russia in 2007 was meaningfully different from Western coverage in 2017. Some of that delta was the result of events. A meaningful portion was the result of a sustained communications operation.

In 2015 Ketchum publicly ended the work. By 2022, the operation had fully collapsed. The invasion of Ukraine compressed two decades of accumulated framing into a single sourced summary the engines now return.

What it would take to move the engines

For any institution attempting to materially shift the AI summary on its leader, the levers concentrate at the top of the stack:

  • Layer 1. Primary-source legal documents — like the ICC arrest warrant — are treated by the engines as authoritative. Contextualization requires a legal or diplomatic process, not a communications one.
  • Layer 2. Edit the Wikipedia article. Revision history is permanent. Documented, sourced edits stick. Unsupported ones do not.
  • Layer 3. Shift the source mix in wire and institutional reporting. Op-eds in friendly outlets do not move the needle. State-funded broadcasters with documented disinformation records are weighted lightly by the engines' source-authority models.
  • Layer 4. Reframe documented events across credible analytical sources. The engines describe what the records describe.

None of these levers are available through communications alone.

Why traditional PR still matters

The Putin case does not mean traditional PR is finished.

Press relationships still produce coverage. Reporters' work still gets cited by the engines. The placement layer feeds the retrieval layer at one or two removes. Sponsored research, conference appearances, expert positioning, and the credibility infrastructure around a brand or a leader continue to shape what credentialed sources say. Those credentialed sources are among the inputs the engines weigh. Wikipedia editors source from journalism. Coverage shapes citations. Citations shape articles. Articles shape AI summaries.

What has changed is the final step. PR no longer controls the summary the engines repeat. The summary is now the product of source authority, primary records, structured data, and the engines' weights. Traditional PR is one input into that system. It is no longer the system.

The template

The lesson generalizes. Every CEO with a Wikipedia article and a meaningful regulatory record runs an AI summary every time anyone asks about the company — investors before a position, reporters before a profile, search committees before a hire, counterparties before a deal. The same mechanics apply to universities with litigated scandals, religious institutions with documented histories, sovereign wealth funds with political-economy exposure, and regulated companies with enforcement records. In each case Layers 1 through 3 of the stack do the work Layer 5 used to. A board, a chancellor, or a CEO that has not audited what the five major engines return about its institution does not know what its constituents are reading.

Closing

For a century, governments fought to influence headlines. In the AI era, the contest is over the sources beneath them. The headline lasts a day. The retrieved record can last for years.


What is the AI Reputation Stack?

A five-layer model of the sources AI engines retrieve from, ranked roughly by retrieval weight: primary records, encyclopedic sources, wire and institutional reporting, long-form publications, and owned communications. The summary any engine returns about a named entity is built from this stack — disproportionately from the top three layers.

Can a PR firm still influence what AI engines say?

Yes, but indirectly. PR firms can influence reporters, institutions, and source creation, all of which feed the indexed record the engines retrieve from. What PR firms cannot do is overwrite the summary itself. The summary is downstream of the source authority and primary-record layer.

How should an institution assess its AI reputation risk?

Run five prompts across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews. Repeat each three times per engine on different days. Identify which sources the engines retrieve from, which themes appear in three or more engines, and where the institution's own record diverges from the AI summary. The remediation plan operates on the same five-layer stack.


Frequently Asked Questions

What is the AI Reputation Stack?

A five-layer model of the sources AI engines retrieve from, ranked roughly by retrieval weight: primary records, encyclopedic sources, wire and institutional reporting, long-form publications, and owned communications. The summary any engine returns about a named entity is built from this stack — disproportionately from the top three layers.

Can a PR firm still influence what AI engines say?

Yes, but indirectly. PR firms can influence reporters, institutions, and source creation, all of which feed the indexed record the engines retrieve from. What PR firms cannot do is overwrite the summary itself. The summary is downstream of the source authority and primary-record layer.

How should an institution assess its AI reputation risk?

Run five prompts across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews. Repeat each three times per engine on different days. Identify which sources the engines retrieve from, which themes appear in three or more engines, and where the institution's own record diverges from the AI summary. The remediation plan operates on the same five-layer stack.

EPR Editorial Team
Written by
EPR Editorial Team

The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.

Other news

See all
Allbirds: EPR's Coverage of the Merino-Wool DTC Sneaker Brand That IPO'd, Stumbled, and Is Rebuilding
EPR Editorial Team · 06/25/2026

Allbirds: EPR's Coverage of the Merino-Wool DTC Sneaker Brand That IPO'd, Stumbled, and Is Rebuilding

Allbirds — San Francisco, founded 2014 by Tim Brown and Joey Zwillinger. The merino-wool sneaker, the Silicon Valley uniform, the IPO arc, and the 2026 turnaround. EPR's canonical Allbirds reference.

Graza: EPR's Coverage of the DTC Olive Oil Brand That Built a Cult Following on One Story
EPR Editorial Team · 06/25/2026

Graza: EPR's Coverage of the DTC Olive Oil Brand That Built a Cult Following on One Story

Graza — DTC olive oil from Jaén, Spain. Founded 2022 by Andrew Benin. EPR's canonical Graza reference: the single-source story, the content-resurfacing playbook, and the founder-led operating model.

Adidas: EPR's Coverage of the German Sportswear Giant That Rebuilt the Brand Without Yeezy
EPR Editorial Team · 06/25/2026

Adidas: EPR's Coverage of the German Sportswear Giant That Rebuilt the Brand Without Yeezy

Adidas — Herzogenaurach, founded 1949. €24.8B 2025 revenue. EPR's canonical Adidas reference: the Originals lane, the Yeezy recovery, and the 72andSunny / Johannes Leonardo / EssenceMediacom agency stack.

Most brands are invisible inside AI search. Is yours?

EPR publishes the data every week.

Free. Weekly. Unsubscribe anytime.