Gard Calls It "AIS-Assisted Collisions" — Four Real Cases Where the Screen Was Trusted Over the Lookout

A Pattern With a Name

Gard has given a recurring casualty pattern an unambiguous name: AIS-assisted collisions. Not AIS failure — AIS working exactly as designed, and a bridge team trusting it more than it deserves. "AIS has evolved into an important safety aid for situational awareness and even collision avoidance," Gard's guidance notes. "However, this evolution also brings new risks, and over-reliance on AIS may contribute to incidents."

The guidance isn't theoretical. It's built around four real collisions, with a combined death toll that should stop any watchkeeper mid-scroll.

Four Cases, One Pattern

6 January 2018, East China Sea, 1950LT — A tanker and a bulk carrier collided. 32 lives were lost aboard the tanker, along with significant pollution. AIS data discrepancies were a key factor — course over ground off by 20–25°, speed over ground off by 2–3 knots.

18 September 2020, Laotieshan Traffic Separation Scheme, 0419LT — A bulk carrier and a fishing vessel collided. 10 lives were lost aboard the fishing vessel. Over-reliance on AIS and ECDIS was identified, and the fishing vessel's own AIS was inoperative right up until the collision — meaning the very system being over-trusted wasn't even transmitting a true picture.

22 December 2022, Yellow Sea, approximately 0150LT — A bulk carrier and a fishing vessel collided. 11 lives were lost aboard the fishing vessel. A dangerous AIS target had been showing on the display well before the collision — and wasn't acted on until the final moments.

July 2024, Subic Bay approach, Philippines — A bulk carrier and two tankers collided, damaging all three vessels. Radar echoes had been visible for more than 40 minutes before the collision. AIS was relied upon instead.

Why AIS Earns This Kind of Trust — and Why That's the Problem

AIS gives you a name, a course, a speed, and an instant CPA/TCPA calculation. Radar plotting takes longer and demands more skill to interpret. It's not hard to see why AIS starts to feel like the more reliable picture, especially in busy or restricted waters where workload is already high.

Gard's guidance is direct about why that trust is misplaced: "AIS data reliability depends on the accuracy of its connected sensors. If AIS transmits inaccurate information, it creates situational confusion rather than improving awareness." And more fundamentally: "Not every vessel has a working AIS, so it is important not to rely only on AIS for lookout."

The specific limitations behind the four cases above:

• Refresh delays in data transmission — the picture on your display is not necessarily the picture right now
• Sensor dependency — AIS is only as accurate as the GPS, gyro and speed log feeding it, and any of those can be wrong without an obvious fault indication
• Incomplete coverage — not every vessel carries AIS, some operate in passive mode, and units can simply be switched off or malfunctioning (as the Laotieshan fishing vessel's was)
• GPS interference — increasingly common in conflict zones, and GPS "smoothing" can affect portable pilot unit accuracy specifically
• Different calculation basis — AIS calculates collision risk from dynamic data broadcast by both vessels; radar plotting evaluates how a target is actually moving relative to your own ship, a genuinely different (and complementary) method

What Gard Recommends

• Maintain a proper lookout using all available means — not AIS alone, not radar alone
• Conduct systematic radar plotting for target vessels, not just a glance at AIS-derived CPA/TCPA
• Use visual observation as the primary cross-check whenever conditions allow it
• Activate AIS targets on the radar display so both data sources are visible together, not toggled between
• Monitor approaching vessels early — the Subic Bay case had over 40 minutes of radar warning that went unused
• Integrate multiple information sources rather than defaulting to whichever one is fastest to read
• Keep terrestrial navigation skills current — manual fixes, radar overlay — so they're not rusty exactly when a sensor picture goes wrong
• Reinforce all of the above through the SMS and standing orders, not just individual good habits

The Uncomfortable Common Thread

None of these four collisions happened because AIS was broken in a way anyone could see. Case 2's fishing vessel had a genuinely inoperative AIS — invisible to the other vessel's screen, not flagged as an error. Case 1's AIS was transmitting, just inaccurately. Cases 3 and 4 had perfectly good information available — a dangerous target, 40 minutes of radar warning — that simply wasn't acted on quickly enough because the AIS picture felt sufficient in the moment.

That's the actual lesson: AIS over-reliance isn't usually a single dramatic failure. It's a slow erosion of the habit of cross-checking, task by task, watch by watch, until the one time it matters most.

Test Your Knowledge

Think you've got this covered? Put it to the test in Crew Connect's free Knowledge Checker — try a quick navigation round and see how you score.

Related Reading

• This week's SRG roundup — where this Gard release was first flagged
• Matthew Flinders III — another case of a working backup being under-trusted, not over-trusted
• Gard's 2026 Crew Claims Report — why routine work injures more crew than emergencies do