How the answers frame BP
In this snapshot from 28 September 2026, BP was named in 9 of 25 answers from ChatGPT, Claude, Gemini, Perplexity and Grok. Shell was named in 10 and ExxonMobil in 8. All nine BP answers came from three questions that never named a company: the worst environmental record among oil and gas companies, the energy companies doing the most on climate or leading the clean transition, and which energy company investors should avoid on ESG grounds. BP appeared in every provider's answer to the worst-record question.
The dominant context was accountability. Claude's answer tied BP to specific past incidents. Perplexity said BP stood out for the largest total environmental penalties in Good Jobs First's Violation Tracker, at about $36.5 billion across 425 records, much of it linked to Deepwater Horizon. Grok's answer cited the same tracker but gave about $25.8 billion across 157 records. Two assistants used the same source family and reported different totals, which is a useful reminder that the figures attached to BP vary by answer.
In the climate-leadership answers, BP sat in a supporting role. Gemini grouped it with Shell, TotalEnergies, Eni and Equinor under "Mixed Progress" and pointed to critics who say low-carbon spending is small next to oil and gas investment. Grok said BP and Shell have renewables exposure and low-carbon targets but generally lag TotalEnergies. In both answers, NextEra Energy and other renewables developers carried the leadership storyline.
The data classes most of BP's credit under Sustainability & Ethics (6 mentions), with Safety & Compliance a distant second (1). That label describes the topic the answers discussed. The answers themselves were largely critical or qualified in tone on that topic, so the topic label should not be read as praise.
…the Texas City refinery explosion, which caused the death of 15 workers, and the 2006 Prudhoe Bay oil spill, the largest oil spill on Alaska's North Slope…
What the press is saying
PeakMetrics' news index shows 17,148 mentions of BP over the last 30 days, and 1,282 of them have BP in the headline. That is heavier than Shell (15,709) and ExxonMobil (12,332) and lighter than Chevron (31,127) in the same sector set.
The five recent headlines supplied are a mixed set. One reports BP starting gas production at Egypt's Fayoum-4 well two years ahead of schedule. One is a stock-momentum piece. The others concern Aker BP's dividend comments, a biopolymer filler product with "BP" in its name, and a garage door company. Some of the volume in this index is therefore likely to come from look-alike names and not from the oil major, and comms teams should check that before reading the total as reputational exposure.
While these companies have made more public-facing commitments to the energy transition, they are still frequently included in "worst" lists by environmental watchdogs.
The gap
The press sample leans toward operations and markets: a production milestone, a momentum-stock call, and coverage of other companies that share the name. The AI answers lean toward history and scorecards: spill and refinery incidents, penalty databases, Oil Change International-style climate ratings, and critics' views on transition spending.
The sources visible in the answers point the same way. ChatGPT cited Carbon Majors and ProPublica material on BP's accident record. Grok cited Good Jobs First and Offshore Technology. Nothing in the supplied answers pointed to BP's own production or investment announcements. That does not show what the assistants weigh internally. It shows which sources appeared in this snapshot.
Other majors like Equinor, Eni, Shell, and BP have renewables exposure and low-carbon targets but generally lag TotalEnergies in pace and scale of renewable buildout…
What to watch
First, the question type matters. When a question asks who is worst or who to avoid, BP was named alongside the same peers, so the comparison set is set by the question. When the question is about leadership, BP was named as a laggard or a mixed case, and renewables pure-plays took the lead.
Second, the numbers attached to BP vary between answers. Comms teams could track which penalty totals and incident references each assistant repeats, and whether they match the underlying databases.
Third, press volume does not carry over into these answers. Fresh operational news, such as the Egypt gas milestone, is the kind of material a team could test for pickup on climate and ESG benchmark questions. This is one snapshot from two runs of five assistants, so repeat the benchmark questions before drawing conclusions.