Corrections

Corrections

What we got wrong, and what each page says now.

About this log

This page lists material errors of fact on TripGuards, newest first. Each date is the day the fix entered our version history. Routine updates, such as new figures or a dated refresh, are not listed.

The log covers corrections we can document from that history, starting in July 2026. It is not a claim that nothing else was ever wrong.

Found an error? Email feedback@tripguards.com.

Guides and pages

7 November 2026 · Dubai, Medellín, Rio de Janeiro, Amsterdam, Phuket, Tokyo, Cancún

What was wrong: Seven guides showed a headline risk score that differed from the City Safety Index by 0.5 or more: Dubai 6 against 5.0, Medellín 6 against 6.7, Rio de Janeiro 6 against 5.3, Amsterdam 3 against 2.4, Phuket 4 against 4.6, Tokyo 3 against 2.5, and Cancún 5 against 4.5.

What it says now: Each guide shows its index score, and its risk label follows the same bands as the city pages. The index itself did not change. The build now fails if a guide drifts 0.5 or more from the index.

6 November 2026 · City pages

What was wrong: City pages showed a risk score written by AI that did not match the City Safety Index: Paris showed 4.5 against an index score of 3.9, and Dubai 3.75 against 5.0. They also said "fact-checked" and gave a last-verified date, which was only the date the page was built.

What it says now: Every index city shows its index score. A page no editor has reviewed now says so, and only a hand-reviewed page carries a review date.

6 November 2026 · Safety guides, Methodology

What was wrong: Said guides were built by analyzing "200+ traveler reports". We found no record behind that count.

What it says now: The count is removed. Guides name what they are built from: government advisories, official statistics where they exist, and local news.

5 November 2026 · Phone theft in London

What was wrong: Called London the highest phone-theft volume of any European city, the worst in Europe and the fastest-improving, without a source for any of those. Said 2026 was the first year the numbers fell.

What it says now: Those claims are gone. Met figures show the fall began in 2025: 81,439 offences in 2024, 71,867 in 2025, and 39,793 in January–August 2026 against 49,136 a year earlier.

5 November 2026 · Is Tokyo safe?

What was wrong: Said Tokyo Metro fares rose in March 2026, placed the Tokyo Medical and Surgical Clinic at Roppongi Hills, and said Japan was willing to detain foreigners with unpaid medical bills. Described a UK advice update of April 28, 2026 as a fresh alert from "days ago".

What it says now: The fare and detention lines are removed: we found no source for them. The clinic is listed at Shiba-koen, Minato. The guide says the UK updated its drink-spiking advice on that date.

5 November 2026 · Is Phuket safe?

What was wrong: Said most visitors get 60 days visa-free, and that travel insurance is void without an International Driving Permit. Quoted cannabis penalties, shop closures and Russian visitor numbers with no source on file.

What it says now: The visa-free stay is 30 days since September 15, 2026. The insurance line follows UK advice: your insurer may not cover you if you are not licensed for the bike. The unsourced figures are removed.

5 November 2026 · Methodology, Disclosure

What was wrong: Described city safety data as "Real-time", fetched fresh on each search, and said priority cities were reviewed for accuracy before publication.

What it says now: City briefings are cached for up to three days, transport and museum briefings for up to seven, and visa data for up to one. The page says most AI briefings are not individually fact-checked.

5 November 2026 · City Safety Index

What was wrong: Recorded Australia's advice for Colombia and Qatar as "Reconsider your need to travel", and for Thailand as normal in the main tourist areas.

What it says now: All three read "Exercise a high degree of caution", taken from Smartraveller on November 5, 2026. No score changed.

5 November 2026 · Pickpocketing in Rome, Pickpocketing statistics

What was wrong: Presented 33,455 reports, about 92 a day and 23.8% of Italy's total, as a figure for the city of Rome.

What it says now: It is labelled as a figure for the Rome metropolitan area. ISTAT's figure for the city is 31,991.

5 November 2026 · Pickpocketing statistics

What was wrong: Said about 4–7% of reported pickpocketing cases are cleared in every European country that publishes a rate.

What it says now: Comparable pickpocketing-only rates run from 2.9% in Italy to 14.3% in Czechia, most under 10%. France is about 7% on a wider category.

5 November 2026 · Is Cancún safe?

What was wrong: Said airport Uber pickups had been legal since a 2024 ruling, wrote an old sargassum forecast as if it were current, and quoted "91/100", "78/100" and "99.9% of Hotel Zone visits incident-free" with no source on file.

What it says now: The guide says federal authorities hold that apps lack airport pickup permits, gives dated sargassum figures, and no longer quotes the three unsourced figures.

2 November 2026 · Is Paris safe?, Pickpocketing in Paris

What was wrong: Said about 18,500 tourist thefts a year, about 150 a day. The figure had no traceable source, and the daily number did not follow from it.

What it says now: The pages use the SSMSI count: about 95,000 non-violent thefts from a person in Paris in 2025, residents and visitors combined. No tourist-only count is published.

2 November 2026 · Is Paris safe?, Paris scams

What was wrong: Gave the CDG taxi flat fares as €53 and €60, with the two banks reversed.

What it says now: €56 to the Right Bank and €65 to the Left Bank.

2 November 2026 · Pickpocketing in Paris, What to do if pickpocketed

What was wrong: Said any RATP ticket desk takes a theft complaint, and described France's online route as a pre-complaint finished with a signature visit.

What it says now: RATP desks take complaints from visitors only, with no violence and an unknown thief. The pré-plainte was replaced on October 15, 2024 by a fully online complaint, available in English.

2 November 2026 · What to do if pickpocketed

What was wrong: Described Berlin as offering online theft reports in English. We could not verify the English part.

What it says now: The claim is removed. The page says to use any station, or the online report when it is running.

2 November 2026 · City Safety Index

What was wrong: Had Australia's advice the wrong way round: Italy as "Exercise a high degree of caution" and the UK as normal precautions.

What it says now: Italy is "Exercise normal safety precautions" (updated October 29, 2026), and the UK is "Exercise a high degree of caution" (updated September 11, 2026). No score changed.

2 November 2026 · Homepage

What was wrong: Advertised "195 countries", "24/7 monitoring", "16 categories" and "updated weekly".

What it says now: The homepage states what the build can check: 50 cities scored, 350 published sub-scores and monthly advisory checks.

2 November 2026 · Is Dubai safe?

What was wrong: Said Dubai had never been hit.

What it says now: The guide records that DXB Terminal 3, the Burj Al Arab and Jebel Ali were struck on March 1, 2026.

3 September 2026 · Is Amsterdam safe?

What was wrong: Said the Netherlands was at US Level 1, "highest safety rating".

What it says now: US Level 2, for terrorism. The check that should have caught this was only matching a fraction of posts, and was rewritten the same day.

4 August 2026 · City Safety Index

What was wrong: The press kit said Doha, Abu Dhabi and Dubai rival Zurich on street crime and all sit in the bottom third on geopolitics. Doha ranked 32nd of 50, outside the bottom third, and all three scored better than Zurich on street crime.

What it says now: The finding says the three share the same street-crime score and advisory premium, and are separated by the political sub-score.

30 July 2026 · City Safety Index

What was wrong: A finding said Istanbul ranked worse than Dubai after the July 27 review had already reversed that pair.

What it says now: The finding was rewritten around the corrected scores: Dubai 5.0, Istanbul 4.9.

30 July 2026 · Is New York safe?

What was wrong: Quoted subway figures that came from news aggregators, not the NYPD or the MTA.

What it says now: The hard figures were removed.

23 July 2026 · Is Rome safe?, Is Barcelona safe?

What was wrong: Said the EU Entry/Exit System had not started, and described Italy and Spain as US Level 1, the "highest safety rating".

What it says now: The EES has been in force since April 10, 2026, and both countries are shown at Level 2.

23 July 2026 · Is Dubai safe?

What was wrong: Showed the US advisory for the UAE as Level 2, and Australia's as Do Not Travel.

What it says now: US Level 3, in effect since March 2026, and Australia at Reconsider your need to travel, as of that date.

City Safety Index changelog

Score changes and advisory re-checks for the City Safety Index, as recorded with the data. Most are re-checks that changed nothing, and those are kept here too.

7 November 2026

Seven guides (Tokyo, Amsterdam, Phuket, Cancún, Dubai, Medellín, Rio de Janeiro) published a rounded editorial score that differed from the index by 0.5 or more. Their headline score now matches the index. No index score, sub-score or advisory moved.

5 November 2026

Advisory re-verification. Every machine-checkable row, now including Denmark and China, matches the live State Department feed, and Israel is unchanged at Level 3 country-wide (feed entry dated 23 Oct 2026). Wording only, no level or score moved: Thailand rows now say Exercise Increased Caution near the Thai-Cambodian border and in Yala, Pattani and Narathiwat, replacing an unsupported "do-not-travel zones"; Indonesia adds unrest and names Central Papua and Highland Papua; Egypt adds border areas (military zones); UAE reason reads terrorism and armed conflict.

2 November 2026

Advisory re-verification. Every machine-checkable row still matches the live State Department feed and no score moved. Checked by hand against the same feed: Denmark Level 2 and China Level 2 (both unchanged; the feed names them in a form the script does not match), and Israel Level 3 country-wide, reissued 23 Oct 2026 with the same Level 4 zones (north of Route 85/90/87, within 11.3 km of Gaza, within 2.4 km of the Egypt border except Taba). Tel Aviv stays outside every Level 4 zone, so its score is unchanged.

8 September 2026

Advisory re-verification. All 45 machine-checkable rows still match the live State Department feed and no score moved. Israel — the one row a machine cannot settle — confirmed by hand: Level 3 country-wide, reissued 26 Aug 2026 with "no changes to the Travel Advisory Level"; Gaza Level 4; the West Bank now published as its own Level 3 entry. Tel Aviv therefore keeps its +1.5 premium: base 3.9, total 5.4, joint 46th. Feed change worth recording: the Israel entry now carries a country-wide level of its own, where until August 2026 it deferred to per-area summaries. check-advisories.mjs now cross-checks that level against the feed instead of skipping the row, and still lists it for a human, because the Level 4 zones are geographic (Gaza, north of Route 85 to Route 90 to Route 87, 11.3 km of the Gaza Strip, 2.4 km of the Egypt border except Taba) and the index scores Tel Aviv, which lies outside all of them.

5 August 2026

First automated sweep of all 50 cities against the State Department advisory feed (scripts/check-advisories.mjs), replacing spot-checks of the adjustment rows alone. No score drift: every acute adjustment still matches its live advisory. Two recorded levels were stale and are corrected — Denmark L1→L2 (Kingdom of Denmark, terrorism; had been Level 2 since 2024) and Malaysia L2→L1 (lowered 2026-02-22; the eastern Sabah kidnapping warning is regional and does not reach KL). Neither crosses the L3 threshold, so no premium and no total changed. Israel is not machine-checkable — its feed entry defers to per-area summaries — and was confirmed by hand at Level 3 country-wide.

30 July 2026

Advisory re-verification of all five acute-adjustment rows (Dubai, Abu Dhabi, Doha, Tel Aviv, Medellín). All four underlying advisories confirmed unchanged: UAE Level 3, Qatar Level 3, Israel Level 3 country-wide (Gaza and border zones Level 4), Colombia Level 3 country-wide with Medellín outside the Level 4 zones. No scores changed.

27 July 2026

Dubai political 9→10 (3.3→3.5 base, 4.8→5.0 total): July re-escalation — ceasefire collapsed Jul 9, Strait of Hormuz closure, US worldwide caution Jul 18. Earlier plan to revise the Dubai blog score 6→5 cancelled for the same reason; 6 stands.

23 July 2026

Edition 2026.1 frozen. 50 cities scored; advisories cross-verified via 3-source check (live web + two independent AI reviews). Post-review changes: Johannesburg crime 9→10 (global top-tier violent crime, review finding); South Africa confirmed US L2 (no adjustment); acute-adjustment rule (+1.5 L3 / +3.0 L4) retained after review — base and adjustment shown separately by design.