All was not well in the automated election system on Election Day 2025, according to the poll watchdog Kontra Daya.

Toward the 7 p.m. close of polling precincts nationwide on Monday, May 12, Kontra Daya said the errors in the automated counting machines (ACM) accounted for most of the problems encountered in the midterm polls.

It said the “verified reports” it had received concerning ACM errors included paper jams, “overvoting” due to “slight smudges,” or markings on the ballot, and mismatches between the actual candidates chosen by voters and the results shown on the receipt of the names on the ballot.

This receipt is called the VVPAT, or the voter verifiable paper audit trail, which is a slip of paper that emerges from the ACM after the ballot is inserted into it and read by the machine. It is like a receipt from a grocery or supermarket cash register.

The VVPAT is an important document in checking for election fraud.

Kontra Daya said overvoting would nullify all the choices made by the voter for a certain position such as senator, and the VVPAT-ballot mismatch would result in a wrong count.

Disenfranchisement

The outcome would be “critical in the sense that it may have resulted in disenfranchisement and inaccurate counting of the votes,” Kontra Daya convenor Danny Arao told CoverStory.

He said the ACM being used in the midterm elections, which was made by a Korean company, is “oversensitive,” so that a small smudge, if it is within the circle to be shaded beside the name of a candidate, could be counted for that candidate even if it is not the voter’s choice.

According to unverified social media reports, some voters had complained that the VVPAT showed names of senatorial candidates they did not vote for.

If the smudge is outside the circle, it could be considered an overvote, and invalidate all the votes for that position. In the case of the Senate race, the machine would report that the ballot contained more votes than the maximum 12 for senators.

Arao said the disenfranchisement due to a machine problem also occurred for the “flimsiest of reasons,” such as the skewed insertion of ballots or when the ACMs do not properly read the votes because of smudges.

Hours before the balloting ended, Kontra Daya reported 305 cases of ACM errors, or 54% percent of verified reports, which include malfunctions that caused hour-long delays.

These delays, Arao said, could also lead to disenfranchisement as some voters may decide to leave without filling out their ballots.

Kontra Daya also received reports of 44 incidents of illegal campaigning, or 44% of Election Day problems. Disenfranchisement (41 incidents or 7%) included missing voters’ names from the list of voters and “procedural barriers.”

There were also 35 incidents of Red-tagging (in which candidates or groups are labeled as supporters if not members of insurgent groups). Noncompliance of rules by members of the Board of Election Inspectors accounted for 34 cases or 6%.

Vote-buying (3%), election-related violence (3%), black propaganda (2%), tampered ballots (2%), harassment of voters (2%) and other Election Day problems (8%) were also reported by Kontra Daya.

According to Arao, the cases of disenfranchisement will impact tightly contested races where every vote counts. In the senatorial elections, that would impact the candidates in the 10th to 14th positions. In local close fights where the candidates are almost evenly matched, it would also have an impact. 

Obviously, for unopposed candidates, disenfranchisement would not matter at all.

Possible election fraud

Kontra Daya and Computer Professionals Union (CPU) also raised concerns about the allegations made by lawyer Harold Respicio, a candidate for vice mayor of Reina Mercedes, Isabela, of possible election fraud due to a mismatch between the software version used by the ACM for Election Day and what had earlier been approved or certified.

Respicio said in a Facebook post on May 10 that the final version 3.5.0, which he said was not officially certified, was different from the certified audited version, 3.4.0, not only in the version number but also in the hash code.

The software is a set of instructions used by the ACM, from the time the machine is turned on, reads the ballots, transmits the results, and puts out its reports until it shuts down.

Pointing to the reported changes in the software cited by Respicio, CPU said version 3.5.0 had a hash code beginning with 0438, while version 3.4.0 starts with 3AD9.

“A different system version could mean that changes were inserted into the source code, altering its entire function or creating, intentionally or unintentionally, a ‘backdoor’ that may be exploited to manipulate election results,” CPU said in a statement on the eve of the polls. “The implications of this discrepancy are profound.”

The Commission on Elections (Comelec) dismissed the allegation as fake news, saying that version 3.4.0 became version 3.5.0 after it was certified.

Without referring to Respicio by name, Comelec Chair George Erwin Garcia said the “candidate” just saw the hash code of the local source code review report that was audited, and that the final version was the “trusted build” of the software. 

Kontra Daya said in a statement that the ACM and its “inner workings should reflect the people’s vote and be subject to public scrutiny to ensure that it will record our vote.” 

CPU said that regardless of the Comelec’s explanations, the change in the software version “completely undermines the whole process of transparency and accountability” and “raises questions about the integrity of the entire electoral process.”