A Search for Radio Technosignatures from Interstellar Object 3I/ATLAS with the Allen Telescope Array

TL;DR

The study investigates whether the interstellar object 3I/ATLAS hosts detectable radio technosignatures by observing with the Allen Telescope Array across 1–9 GHz for 7.25 hours. A multi-stage pipeline (bliss for Doppler-drifting narrowband hits, RFI blanking, and NBeamAnalysis spatial filtering) reduces an initial ~74 million hits to 211 candidate plots, all of which are attributed to RFI upon visual inspection. No technosignature signals are found, enabling EIRP upper limits of $10$–$110$ W over the surveyed frequency and drift-rate ranges. The work demonstrates drift-rate-aware filtering and a comprehensive end-to-end search pipeline for ISOs, and it motivates future observations (e.g., December 2025) to tighten constraints and extend methods to broader parameter spaces.

Abstract

In 2025 July, the third-ever interstellar object, 3I/ATLAS, was discovered on its ingress into the Solar System. Similar to the NASA Voyager missions sent in 1977, science probes by extraterrestrial life (artifact ``technosignatures'') could be sent to explore other stellar systems like our own. In this campaign, we used the SETI Institute's Allen Telescope Array to observe 3I/ATLAS from 1--9~GHz. We detected nearly 74 million narrowband hits in 7.25~hr of data using the newly-developed search pipeline \texttt{bliss}. We then applied blanking in frequency and drift rate to mitigate Radio Frequency Interference (RFI) in our dataset, narrowing the dataset down to $\sim$2 million hits. These hits were further filtered by the localization code \texttt{NBeamAnalysis}, and the remaining 211 hits were visually inspected in the time-frequency domain. We did not find any signals worthy of additional follow-up. Accounting for the Doppler drift correction and given the non-detection, we are able to set an Effective Isotropic Radiated Power (EIRP) upper limit of $10-110$~W on radio technosignatures from 3I/ATLAS across the frequency and drift rate ranges covered by our survey.

Figures (6)

• Figure 1: The radial acceleration of 3I/ATLAS over the $9$ days containing our observations (blue line). Observation times are highlighted in yellow. This allows us to determine the expected drift rates for potential signals coming from 3I/ATLAS.
• Figure 2: Frequency distribution of the $\sim 74$ million hits obtained in this survey. The top, middle, and bottom panels correspond to the "low" (1000--3688 MHz), "mid" (3688--6376 MHz), and "high" (6376--9064 MHz) frequency ranges, respectively. All y-axes are scaled to the same limits. Hits are shown in blue, blanking ranges are shown with yellow bars, the hits after blanking within the ranges are shown in brown, and the hits after limiting the drift rate range are shown in gold. Applying blanking ranges and drift rate limits significantly reduced dense clusters of hits (reducing the total number of hits by 97.4%) which improved our ability to process the data and detect putative signals associated with 3I/ATLAS.
• Figure 3: The distribution of the $\sim 74$ million hits from this survey across drift rate. Note the peak of hits around 0 Hz/s (expected because most RFI transmitters are in the same reference frame as the telescope, i.e., on the ground). After the drift rate cut, the count of hits is much more even across the restricted drift rate range. Note that the sawtooth pattern seen across drift rate is a known artifact in the current development version of bliss, caused by an overestimation of SNR for high-bandwidth ($\gtrsim 10$ Hz wide) hits close to integer multiples of the unit drift rate. It causes increased amounts of false positives close to these drift rates but does not cause false negatives jacobson-bell2025bliss.
• Figure 4: The distribution of the $\sim 74$ million hits from this survey across SNR. As expected, there are significantly more weak hits than strong ones, and there is no significant trend in SNR when filters are applied to frequency and drift rate.
• Figure 5: An output waterfall plot from NBeamAnalysis for a signal which was ranked in the top 211 events. The on-beam (pointed at 3I/ATLAS) is shown in the left subplot, while the off-beam is shown in the right subplot. Each subplot is a normalized waterfall plot with frequency on the horizontal axis, time on the vertical axis, and intensity (scaled to the on-beam plot from 0--1) indicated by the color bar. This hit had a frequency outside of the blanking ranges, a drift rate in the allowable range, a relatively low DOT score (indicating significant difference between the plots), and an SNR ratio above 5.29 (indicating consistency with a point source on the sky). However, two things immediately discount this signal as an ETI technosignature: (1) the signal in the on-beam is not a narrowband technosignature despite the bliss hit, but rather some kind of broader-band modulated signal, and (2) this frequency is consistent with the L-band downlink for the Iridium satellites. The Iridium RFI explanation is consistent with the point source characteristics of the hit as well, as satellites and other distant human transmitters can mimic the characteristics that we would expect from a true ETI technosignature.