How the answers frame ExxonMobil
This is a single snapshot taken on 2026-09-28 across five assistants (ChatGPT, Claude, Gemini, Perplexity and Grok) and 25 answers to category questions that never named a brand. ExxonMobil appeared in eight of them. Every mention came from one of two questions: "Which oil and gas companies have the worst environmental record?" and "Which energy company should investors avoid on ESG grounds?" All eight were tagged under the same credit theme, Sustainability & Ethics.
On the environmental-record question, all five assistants named ExxonMobil together with Chevron, BP and Shell. The framing varied by answer. ChatGPT placed it third in a short list and described it as a "major climate emitter plus substantial environmental enforcement history." Claude said it had faced criticism for activities that speed up climate change and for denial of global warming. Gemini grouped it with Chevron as U.S.-based majors criticized for "Grossly Insufficient" climate plans. Perplexity and Grok pointed to penalty counts in Good Jobs First's Violation Tracker and similar sources.
On the ESG question, three assistants named it. ChatGPT offered it as "a frequent example" of a company ESG-focused investors screen out, and added that no company is universally the worst. Perplexity called ExxonMobil "the strongest answer from the supplied sources" for a single company to avoid. Grok declined to name a single company and listed ExxonMobil in a group of investor-owned majors, next to state-owned producers and coal-heavy firms. Claude and Gemini did not name ExxonMobil on this question.
Among energy peers in this snapshot, Shell was named in 10 answers, BP in 9, ExxonMobil in 8, Chevron and NextEra Energy in 7 each, and Duke Energy in 1.
…major climate emitter plus substantial environmental enforcement history…
What the press is saying
PeakMetrics' news index shows 12,332 mentions of ExxonMobil in the last 30 days, with the brand in the headline of 1,130. That is below Chevron (31,127), BP (17,148) and Shell (15,709) in the same peer set.
The recent headlines sampled are mostly financial and operational. They cover the pricing and results of cash tender offers for senior notes, two pieces asking whether XOM is underperforming the energy sector, ExxonMobil becoming operator of an Azerbaijan unconventional hydrocarbons project, and $15,000 in grants to first responders in Yazoo County. None of the five sampled headlines is about emissions, penalties or ESG ratings.
There is no single official ranking of the “worst” oil and gas companies…
The gap
The press storyline in this sample is capital markets, share performance, new projects and community giving. The AI storyline is environmental accountability. The assistants drew it from sources such as Carbon Majors, Good Jobs First's Violation Tracker, Oil Change International and ESG rating providers, not from the recent news flow.
That means recent coverage and AI framing are measuring different things. The headlines sampled here would not, on their own, predict the reputational context in which these assistants place the company. The gap is also uneven across peers. Chevron has more than twice ExxonMobil's news volume and was named in fewer answers.
What to watch
For a comms team, the takeaway from this snapshot is that these assistants reach for ExxonMobil when a question invites a ranking of environmental blame or ESG avoidance. They did not name it in the other 17 answers in this set. Both questions are ones stakeholders such as investors, journalists and campaigners are likely to ask.
Look at the sources behind the answers. Cited domains in these responses included Carbon Majors, Good Jobs First, Greenpeace, ProPublica and Sustainalytics. These are third-party datasets and advocacy reports, not company or news-wire material. Some answers also hedged, with ChatGPT and Grok both stating that no single company is universally the worst. Those caveats are worth tracking alongside the named examples.
This is one run on one date. Live answers vary between runs, so repeat the benchmark questions before reading anything into the pattern.
