Data update: Riverside County CA SO (Jan 2022 – Aug 2026)

Riverside County is in Southern California. About 2.4 million people lived there at the 2020 census. We loaded two sets of Flock audit records from the Riverside County Sheriff’s Office. The org audit lists searches run by the Sheriff’s Flock users. The network audit lists searches by any agency that reached the Sheriff’s cameras. Both run from January 1, 2022 to mid-August 2026. Both skip December 24–31, 2025. All times below are Pacific time. We wrote about the road permits for the Sheriff’s cameras in No Permit, No Problem. Thanks to DHS Concerned Citizen for obtaining and sharing these records.

The files hold fewer columns than a standard Flock audit export. They have no ID, License Plate, Case #, Filters, or Search Type columns. The network files also have no Name column, so they do not show who ran a search. The 2026 files drop Total Devices Searched as well. The Reason column reads “Investigation” on all but 29,752 records across both audits. Those read “Ivestigation”, “investigation”, or “INVESTIGATION”.

#Org audit metrics

• Type: Organization audit
• Timespan: Jan 1, 2022 – Aug 13, 2026 (1,686 days; Feb 1, 2022 and Dec 24–31, 2025 not covered)
• Records: 2,440,287
• Search types: not in the files
• Operator name: blank on 181,648 records
• License plate and Case #: not in the files
• Reason: “Investigation” on 2,411,704 records; “Ivestigation” on 28,583

The org name on all records is Riverside County CA SO. The org audit has no records for February 1, 2022. The network audit shows 1,042 Riverside County CA SO searches that day. The last org audit record is from 10:42 a.m. on August 13, 2026.

Total Networks Searched reads 0 on every org audit record before 4 p.m. Pacific on December 31, 2022, which is midnight UTC on January 1, 2023. Most records list a count from then on. The median org search covered 149 networks in 2023, 319 in 2024, 870 in 2025, and 938 in 2026.

#Network audit metrics

• Type: Network audit
• Timespan: Jan 1, 2022 – Aug 14, 2026 (1,687 days; Dec 24–31, 2025 not covered)
• Records: 19,919,381
• Organizations (network): 4,622
• Search types: not in the files
• Operator name, license plate, and Case #: not in the files
• Reason: “Investigation” on 19,918,212 records; “investigation” on 1,168; “INVESTIGATION” on 1

The Sheriff’s own searches make up 2,405,858 network audit records (12.1%). The last network audit record is from 10:41 a.m. on August 14, 2026. One network file ends at 8:57 a.m. on May 27, 2026, and the next starts at midnight on May 28. The Anomalies section describes it.

In 2022, Total Networks Searched reads 0 on 1,658,292 of 1,659,532 network audit records. The median network search covered 149 networks in 2023, 489 in 2024, 578 in 2025, and 633 in 2026.

Where two released files cover the same hours, the counts above take those rows from one file only.

The bars show network audit searches per week. The line shows org audit searches per week. Weeks start on Monday. The chart runs from the week of January 3, 2022 to the week of August 3, 2026. It leaves the weeks of December 22 and December 29, 2025 blank. It also leaves the org line blank for the week of January 31, 2022, and the network bar blank for the week of May 25, 2026. The chart counts each exact duplicate row in a file once. The Anomalies section explains the duplicates.

#Out-of-state searches

Agencies outside California searched the Sheriff’s cameras until November 2024. After that month, the network audit shows almost none.

From January 2022 through November 2024, agencies whose Flock names carry another state’s abbreviation ran 3,328,645 of 9,777,512 network searches (34.0%). There are 4,126 such agency names, from 46 states and the District of Columbia. In November 2024, 2,842 of them ran 250,398 searches, 41.5% of that month’s total. In December 2024, three ran 604.

From December 2024 on, three out-of-state agencies ran 3,648 searches. Prescott AZ PD ran 2,461 and San Luis AZ PD ran 899, the last of them in March 2025. An agency named “18th Judicial District DA’s Office- LA PD” ran 288, the last in October 2025. The network audit shows no out-of-state searches after October 2025.

Out-of-state searches also paused once before. On June 5, 2023, out-of-state agencies ran 1,400 searches. From June 6 to August 14, 2023, they ran 1,867 in total, and no more than 195 on any day. On August 15 they ran 995, and on August 16, 3,215.

Houston TX PD ran the most out-of-state searches: 191,528, from January 2022 through November 2024. Dallas TX PD ran 88,857.

Agencies whose names start with “[Federal]” ran 88,170 searches under 14 names. The US Postal Inspection Service ran 56,915, from May 2022 to July 2024. “[Federal] FBI [Inactive]” ran 21,112, the last in July 2023. After November 2024, the only “[Federal]” name in the network audit is the Loma Linda Healthcare System CA Veterans Affairs PD, with two searches in May 2025. No agency name in the network audit refers to ICE, CBP, the Border Patrol, or Homeland Security Investigations.

The California Out-of-State Queries report lists these searches once the data is loaded.

#Top agencies using the Sheriff’s network

#Top operators

The org audit’s Name column holds 2,244 distinct names. Most read “Last, First”. The column is blank on 181,648 records (7.4%).

#Reasons

The reasons in these files say nothing about why anyone searched.

#Org audit

“Ivestigation” appears on all 28,583 records in the November 2023 org sheet. No other reason appears in that sheet. In the May 2023 org sheet, the header cell over the reason column also reads “Investigation”.

#Network audit

Audit logs from other agencies, already on HaveIBeenFlocked, record different reasons for the Sheriff’s searches. For searches dated March 1–7, 2025 (UTC), those logs hold 51,481 Riverside County CA SO records. The reason is blank on 15,182 and “redacted” on 6,961. The rest carry 695 distinct reasons, such as “inv”, “10851”, “hotsheet”, and “stolen”. “Investigation” appears on 798.

#How we processed the files

#Source files

The source is 46 Excel workbooks and seven CSV files.

The workbooks fall into these groups:

• 2022–2024 network workbooks (Network_Audit_<Mon>_<Year>_Final.xlsx): one data sheet each, named like 4_1_2022-4_30_2022-Riverside Co, plus an empty sheet named Sheet1 that holds no values. Six columns: Org Name, Total Networks Searched, Total Devices Searched, Time Frame, Reason, Search Time.
• Jan_-_Dec_2025_Network_Audit_Final.xlsx: 12 sheets named like Jan 2025 Network and Sept 2025 Network, with the same six columns. The Dec 2025 Network sheet ends on December 23.
• 1_1_2026-1_25_2026-Riverside_County_CA_SO-Network-Audit_REDACTED.xlsx and 7_21_2026-7_27_2026-Riverside_County_CA_SO-Network-Audit_REDACTED.xlsx: one sheet each, named in. 7_15_2026-8_14_2026_Network-Audit_Final.csv.xlsx: one sheet. These three have five columns, without Total Devices Searched.
• Org workbooks for 2022, 2023, 2024, and 2025: one sheet per month, named like Org Audit Jan 2022, January 2024, and 1_1_2025-1_31_2025-Riverside Co. Seven columns: Name, then the six network columns. The 2025 workbook also holds an empty Sheet1. Its December sheet ends on December 23.
• Org_Jan_1-Jul_26_2026.xlsx: three sheets. The sheet names do not match their contents. Sheet Jan-Mar 26 holds January, February, March, and May 2026. Sheet Apr-Jun 26 holds April 1 to June 23. Sheet Jun-Jul 26 holds June 23 to July 26. Org_Audit_7_15_2026-8_13_2026_Final.csv_Redacted.xlsx holds one sheet. The 2026 org sheets have six columns, without Total Devices Searched.
• The seven CSV files hold the five-column network layout. In 3_27_2026-4_27_2026-Riverside_County_CA_SO-Network-Audit_REDACTED.csv, the header and every row end with an extra empty field.
• Header cells start with a space, such as " Org Name". In sheet Org Audit May 2023, the reason header reads “Investigation”.

The output table below lists the row count of each sheet we converted.

#Conversion

We converted the workbook sheets with packages/data-load/data/audits/riverside-county-ca/unprocessed/riverside-county-ca-xlsx2csv.py (SHA-256 eb28b66a84494b9631128f0b98131b82f61dedc9fbbcb6c1c47eb7fcf3f7ba5e) under Python 3.12.15 and openpyxl 3.1.5. It adapts the Prescott converter. It converts one sheet per run and takes the sheet name and output path as arguments. The driver riverside-county-ca-stage.sh (SHA-256 c8b2d61db25bbeb8d8a38e81d55a60a0232845d25e4e0ae2dfb87db0d0599110) runs it once for each line of riverside-county-ca-jobs.tsv (SHA-256 fd8a19543baa35571c47d428c0f791dfbbe0b1e31e782eaeb5917f7873588809), which maps each workbook and sheet to a staged file name:

python riverside-county-ca-xlsx2csv.py "<workbook>.xlsx" "<sheet>" <output.csv>

The script opens the workbook with load_workbook(read_only=True, data_only=True). It reads cell values, not formulas. It behaves as follows:

• The first non-empty row is the header. Header text is written as-is, including leading spaces. Trailing empty header cells are dropped.
• A row is empty if every cell is empty. Empty rows are skipped.
• Text cells are written as stored, untrimmed. NUL bytes would be removed; the files hold none.
• Number cells in Total Networks Searched and Total Devices Searched are written as integers.
• Number cells in other columns are written as their stored value. Cells stored as dates, times, or booleans, and numbers with a date, fraction, or other display format, would be written as Excel shows them. The converter found no such cells in these workbooks.
• Empty cells are written as empty strings. Trailing empty cells are trimmed, then the row is padded to the header’s width. Cells beyond the header’s width are dropped.
• Output is UTF-8 without a byte order mark. Every field is quoted. Lines end with LF. Line breaks inside a cell, as in Time Frame, stay inside the quotes.

The script does not parse or change times. Search Time and Time Frame stay as the UTC strings in the source, such as “01/31/2026, 03:57:12 AM UTC”. The conversion is deterministic.

We copied the seven CSV files without changing their bytes and renamed them by date range.

#Changes

These are all the changes, in order:

1. Empty rows skipped during conversion. All sit at the end of their sheets. Network sheets: Jan 2022 row 43936, Feb 2022 row 50312, Mar 2022 row 68563, Apr 2022 row 77166, Nov 2022 row 225158, Jan 2023 rows 181445–181446, Feb 2023 rows 143176–143177, Mar 2023 row 219938, Apr 2023 rows 188041–188042, May 2023 rows 220177–220178, Jun 2023 row 148417, Sep 2023 row 260716, Dec 2023 row 295587, Mar 2024 row 411117, and Apr 2024 row 424335. Org sheet Jan-Mar 26: row 213262.
2. Stray cell dropped during conversion. Sheet Org Audit Feb 2023, row 21, column I holds the text "Org ". The header has seven columns. The rest of row 21 matches the rows around it.
3. Header rename. We then ran packages/data-load/data/audits/riverside-county-ca/unprocessed/riverside-county-ca-fixups.py (SHA-256 caf8c90d814bc0e12f57fcf1fa9fbef1a043461f1efb348c222b81c898d98d9f) from the staging directory. In 2023-05-Organization-Audit.csv, it renames the sixth header cell from “Investigation” to " Reason". The importer reads the reason from a column named “Reason”. No data cell changed.
4. Row dropped. In 2022-01-Network-Audit.csv, the same script drops the last data row. It comes from Excel row 43935 of the Jan 2022 network sheet. Its only content is a single space in the Reason column. It has no Search Time and no Org Name, so the importer would have filed it as a Riverside County CA SO search with no time.

We did not add or remove any column, trim values, change time zones, or split or merge files.

We did not stage two workbooks. 7_15_2026-8_14_2026_Network-Audit_Final.csv.xlsx holds 590,536 rows. They are the same rows as 7_15_2026-8_14_2026_Network-Audit_Redacted.csv. 7_21_2026-7_27_2026-Riverside_County_CA_SO-Network-Audit_REDACTED.xlsx holds 130,958 rows. All of them appear in 7_15_2026-8_14_2026_Network-Audit_Redacted.csv.

#Output

“Sheet rows” counts the rows below the header, including empty rows. For the CSV files it counts CSV records. The SHA-256 values apply to the files after all changes above. We re-ran the converter, job list, and fix-up script on the released workbooks in a separate directory. All 101 converted files matched the staged files byte for byte. The seven copied CSV files have the same SHA-256 as the released files.

#Checked and left unchanged

• No workbook sheet reaches Excel’s 1,048,576-row limit. The largest, the October 2024 network sheet, has 628,968 rows including the header. Two CSV files have exactly 1,048,575 records below the header. The Anomalies section describes them.
• Some released files cover the same hours. The network CSVs for March 27 – April 27 and April 27 – May 27, 2026 both hold April 27. The network CSVs for May 28 – June 23 and June 23 – July 20 both hold part of June 23. The network CSVs for June 23 – July 20 and July 15 – August 14 both hold July 15 to 12:00 p.m. on July 20. In each case, the two files hold the same rows for the hours they share. The org sheets Apr-Jun 26 and Jun-Jul 26 hold the same rows for the hours of June 23 that both cover. The Jun-Jul 26 sheet and the July 15 – August 13 workbook hold the same distinct rows for July 15–26. We left all of them in place. The importer stores one copy of identical rows.
• We kept the Reason values as they appear, including “Ivestigation”.
• Total Devices Searched holds “——” on many 2022–2025 rows. The importer reads it as empty.
• Some Org Name values end with a space, such as "Hemet CA PD ". The importer trims them.
• Many rows exactly match another row in the same file. The files have no IDs, so they do not show whether the copies are repeated searches or repeated rows. We kept all of them.
• Every Search Time in the staged files parses as a UTC timestamp.

#Anomalies

#Repeated rows

Many rows in both audits repeat exactly. Most of them fall between August 2022 and January 2023. In those months, most rows appear in groups of two or more identical copies, most often two. The February 2025 network file has a second cluster, with many rows repeated five times. The files have no ID column, so they cannot show whether a repeat is a second search or a copy of the same row. The importer keeps one copy of each set of identical rows. Three examples, with times in UTC as in the files:

#Excel row limit

Two network CSV files have exactly 1,048,575 records below the header, which is the most an Excel sheet can hold. In 4_27_2026-5_27_2026-Riverside_County_CA_SO-Network-Audit_REDACTED.csv, 568,932 rows hold data and the rest are empty. In 6_23_2026-7_20_2026-Riverside_County_CA_SO-Network-Audit_REDACTED.csv, 532,660 rows hold data and the rest are empty. The empty rows follow the data. The first file’s last search is at 8:57 a.m. on May 27, 2026. The next file starts at midnight on May 28. May 27 has 4,833 network audit records. The days around it have about 21,000 each.

#Missing days

Neither audit covers December 24–31, 2025. The 2025 workbooks end on December 23, and the 2026 files start on January 1. The org audit also skips February 1, 2022.